JP2021536699A - Motion vector acquisition methods, devices, computer devices and storage media - Google Patents

Abstract

The present application discloses motion vector acquisition methods and devices, computer devices, and storage media, and relates to the field of video compression technology. In the method, the initial motion vector of the picture block to be processed is determined by using the positional relationship between the reference block and the picture block to be processed. If the reference block and the picture block to be processed are located in the same coding tree block, the decoder uses the initial motion vector of the reference block as the initial motion vector of the picture block to be processed. .. If the reference block and the picture block to be processed are located in different coding tree blocks, the decoder uses the final motion vector of the reference block as the initial motion vector of the picture block to be processed. do. In this case, if the final motion vector of the reference block is needed for the picture block to be processed, the initial motion vector of the reference block can be used so that the picture block to be processed can be used. It can be used as the final motion vector of the reference block, which means that the picture block to be processed can be decoded only after the final motion vector of the reference block is obtained. Cases can be avoided and decryption efficiency can be improved.

Description

This application claims priority to Chinese Patent Application No. 201811020181.9, entitled "Video Coding Methods and Devices" filed September 3, 2018, which is incorporated herein by reference in its entirety.

This application claims priority to Chinese Patent Application No. 201811271726.3 entitled "Motion Vector Acquisition Methods and Devices, Computer Devices, and Storage Media" filed October 29, 2018, in its entirety. Is incorporated herein by reference.

Technical Fields The present application relates to the field of video compression technology, in particular motion vector acquisition methods and devices, computer devices, and storage media.

Background Video-related applications are becoming more and more popular in everyday life. With the maturity of computer technology, video processing technology has also evolved significantly. Video coding technology has evolved tremendously. Redundancy can be eliminated as much as possible by utilizing the correlation between the coding unit and the prediction unit when the coding unit within the frame of the picture is encoded. At the time of decoding, the information acquired after the redundancy is removed is decoded to acquire the picture information corresponding to the coding unit.

The decoding process may be as follows: After acquiring the motion information of the coding unit, the decoder sets a list of candidate predicted motion vectors based on the motion information and the acquired motion. Coding by selecting the optimal predicted motion vector from the list of candidate predicted motion vectors based on the information and using the reference unit indicated by the optimal predicted motion vector as the starting point. Search for the prediction unit that most closely resembles the unit and update the best predicted motion vector in the list of candidate predicted motion vectors to the motion vector of the prediction unit (ie, the updated predicted motion). The vector is the predicted motion vector of the coding unit), and the picture information corresponding to the coding unit is acquired based on the updated predicted motion vector and the prediction unit.

In the process of realizing the present application, the inventor has found that the related techniques have at least the following problems: the decoder may decode multiple coding units simultaneously during decoding. However, if the predicted motion vector of the current coding unit needs to be used by another coding unit, decoding will update the optimal predicted motion vector of the current coding unit. It can only be executed after it has been done, and as a result, delays are very likely to occur.

Embodiments of the present application provide motion vector acquisition methods and devices, computer devices, and storage media to solve the decoding delay problem in related arts.

According to one aspect, the present application provides a motion vector acquisition method, which is:
A step that determines the reference block of the picture block to be processed, in which the reference block and the picture block to be processed are located in the same frame of the picture.
If the reference block is within a preset range, it is a step to get the motion vector of the picture block to be processed based on the initial motion vector of the reference block, and the preset range is processed. Steps and steps that are determined based on the position of the target picture block,
This is the step to get the motion vector of the picture block to be processed based on the final motion vector of the reference block when the reference block is out of the preset range, and the final motion vector is the initial. Includes steps, which are obtained based on the motion vector of.

In a possible implementation, the fact that the reference block is within the preset range means that the coding tree block CTB where the reference block is located and the coding tree block where the picture block to be processed is located are on the same line. Including being located in
Correspondingly, the fact that the reference block is out of the preset range means that the coding tree block where the reference block is located and the coding tree block where the picture block to be processed is located are located on different lines. Including doing.

Based on the possible implementations described above, different preset ranges may be provided so that the motion vector of the picture block to be processed is determined based on the different preset ranges. It is possible to select multiple motion vectors of the picture block to be processed.

In a possible implementation, the coding tree block where the reference block is located and the coding tree block where the picture block to be processed is located are on different lines, and the coding tree block where the reference block is located is , Above or in the upper left of the coding tree block where the picture block to be processed is located.

Based on the possible implementations described above, it is possible to provide another pre-set range so that multiple pre-set conditions can be provided for motion vector selection.

In a possible implementation, the fact that the reference block is within the preset range includes that the reference block and the picture block to be processed are located in the same coding tree block.
Correspondingly, the fact that the reference block is out of the preset range includes that the reference block and the picture block to be processed are located in different coding tree blocks.

Multiple possible specific pre-set ranges may be provided based on the possible implementations described above.

In a possible implementation, the fact that the reference block is within the preset range means that the coding tree block in which the reference block is located is the same as the coding tree block in which the picture block to be processed is located. , Or the coding tree block in which the reference block is located is a left-adjacent or right-adjacent block of the coding tree block in which the picture block to be processed is located, and correspondingly, the reference block is pre-located. Being out of the set range means that the coding tree block in which the reference block is located is not the same as the coding tree block in which the picture block to be processed is located, or the coding tree block in which the reference block is located. Includes that the tree block is not the left or right adjacency block of the coding tree block in which the picture block to be processed is located.

Based on the possible implementations described above, a plurality of possible specific pre-set ranges may be provided.

In a possible implementation, the step of determining the reference block of the picture block to be processed is:
A step of continuously determining one or more pre-set candidate reference blocks as reference blocks in a pre-set order, where the candidate reference block is a pre-set space with the picture block to be processed. Includes steps, including picture blocks that have a relative positional relationship.

Based on the above implementation, the reference block of the picture block to be processed may be determined, and then the initial motion vector of the picture block to be processed may be determined based on the reference block.

In a possible implementation, the step of determining the reference block of the picture block to be processed is:
The step of parsing the bitstream to obtain one or more first identification information,
A step of determining a reference block from a plurality of candidate reference blocks of a picture block to be processed based on one or more first identification information, wherein the candidate reference block is in advance with the picture block to be processed. Includes steps, including picture blocks with set spatial positional relationships.

Based on the possible implementation described above, the reference block of the picture block to be processed can be determined based on the identification information in the bitstream, and then the initial motion vector of the picture block to be processed. Can be determined based on the reference block.

In a possible implementation, it is possible that the final motion vector is obtained based on the initial motion vector.
To get multiple candidates for the final motion vector by adding the initial motion vector and multiple preset offset vectors separately,
This includes determining the final motion vector candidate corresponding to the minimum distortion cost among the plurality of final motion vector candidates as the final motion vector.

Based on the possible implementations described above, the final motion vector candidate corresponding to the lowest strain cost is used as the final motion vector, resulting in a more accurate final motion vector.

In a possible implementation, the method is used for bidirectional inter-prediction, the final motion vector contains a first final motion vector and a second final motion vector, and the initial motion vector is the first. The initial motion vector and the second initial motion vector are included, and the first final motion vector and the first initial motion vector are based on the first reference frame list of the picture block to be processed. The second final motion vector and the second initial motion vector indicate the motion compensation block based on the second reference frame list of the picture block to be processed.
That the final motion vector is obtained based on the initial motion vector,
To obtain multiple candidates for the first final motion vector by adding the first initial motion vector and multiple preset offset vectors separately.
The candidate for the first final motion vector corresponding to the minimum strain cost among the plurality of candidates for the first final motion vector is determined as the first final motion vector. The first final motion vector corresponds to the first offset vector in a plurality of preset offset vectors.
To obtain the second offset vector, the size of the second offset vector is equal to that of the first offset vector, and the direction of the second offset vector is opposite to that of the first offset vector. There is, that,
It involves adding a second initial motion vector and a second offset vector to obtain a second final motion vector.

Based on the possible implementations described above, the final motion vector is obtained based on the initial motion vector in bidirectional prediction mode.

In a possible implementation, the method is used for bidirectional inter-prediction, the final motion vector contains a first final motion vector and a second final motion vector, and the initial motion vector is the first. The initial motion vector and the second initial motion vector are included, and the first final motion vector and the first initial motion vector are based on the first reference frame list of the picture block to be processed. The second final motion vector and the second initial motion vector indicate the motion compensation block based on the second reference frame list of the picture block to be processed.
That the final motion vector is obtained based on the initial motion vector,
To obtain multiple candidates for the first final motion vector by adding the first initial motion vector and multiple preset offset vectors separately.
Determining the candidate for the first final motion vector corresponding to the minimum strain cost among the plurality of candidates for the first final motion vector as the first final motion vector,
To obtain multiple candidates for the second final motion vector by adding the second initial motion vector and the plurality of pre-set second offset vectors separately.
This includes determining the candidate for the second final motion vector corresponding to the minimum strain cost among the plurality of candidates for the second final motion vector as the second final motion vector.

Based on the above possible implementation, another method of obtaining the final motion vector based on the initial motion vector in the bidirectional prediction mode is provided, resulting in the final motion vector in the bidirectional prediction mode. It is possible to provide multiple ways to obtain.

According to the second aspect, the present application provides a method for determining a motion vector residual, which is:
The step of analyzing the bitstream to obtain the second identification information, which is used to determine the initial motion vector of the picture block to be processed.
The step of adding the initial motion vector and multiple pre-set offset vectors separately to get multiple candidates for the final motion vector,
A step of determining the final motion vector candidate corresponding to the minimum distortion cost among multiple candidates of the final motion vector as the final motion vector, and
Either the difference between the final motion vector and the initial motion vector is used as the motion vector residual of the picture block to be processed, or the final motion vector is the motion vector of the picture block to be processed. Includes steps to use as residuals.

In a possible implementation, the method is used for bidirectional inter-prediction, the final motion vector contains a first final motion vector and a second final motion vector, and the initial motion vector is the first. The initial motion vector and the second initial motion vector are included, and the first final motion vector and the first initial motion vector are based on the first reference frame list of the picture block to be processed. The second final motion vector and the second initial motion vector indicate the motion compensation block based on the second reference frame list of the picture block to be processed.
The step of using the difference between the final motion vector and the initial motion vector as the motion vector residuals of the picture block to be processed is:
It comprises the step of using the difference between the first final motion vector and the first initial motion vector as the first motion vector residual of the picture block to be processed.

Based on the possible implementation described above, the motion vector residuals of the picture block to be processed can be determined in the bidirectional interprediction mode.

In a possible implementation, the step of using the difference between the final motion vector and the initial motion vector as the motion vector residuals of the picture block being processed is:
Further included is the step of using the difference between the second final motion vector and the second initial motion vector as the second motion vector residual of the picture block to be processed.

In a possible implementation, the step of using the final motion vector as the motion vector residuals of the picture block to be processed is:
It comprises the step of using the first final motion vector as the first motion vector residual of the picture block to be processed.

In a possible implementation, the step of using the final motion vector as the motion vector residuals of the picture block to be processed is:
It further comprises the step of using the second final motion vector as the second motion vector residual of the picture block to be processed.

Based on the multiple possible implementations described above, multiple methods are provided to determine the motion vector residuals of the picture block to be processed so that the encoder and decoder have coding efficiency based on the motion vector residuals. And the decoding efficiency can be improved.

According to a third aspect, the present application provides a motion vector data storage method, which is:
The step of analyzing the bitstream to obtain the second and third identification information, which is used to determine the initial predicted motion vector of the picture block to be processed. To be done, steps and
Steps to get the final predicted motion vector based on the initial predicted motion vector and multiple pre-set offset vectors,
When the third identification information indicates that the bitstream carries the motion vector residuals of the picture block to be processed, the bitstream is analyzed to obtain the motion vector residuals, and the motion vector residuals are obtained. Steps to store in the target storage space and
If the third identification information indicates that the bitstream does not carry the motion vector residuals of the picture block being processed, then between the final predicted motion vector and the initial predicted motion vector. Includes the steps of storing the differences in the target storage space or storing the final predicted motion vector in the target storage space.

In a possible implementation, the third identification information indicates that the bitstream carries the motion vector residuals of the picture block to be processed.
The third identification information includes indicating that the prediction mode of the picture block to be processed is the AMVP mode.

Based on the possible implementation described above, the motion vector data of the picture block to be processed can be stored in multiple prediction modes.

In a possible implementation, the third identification information indicates that the bitstream does not carry the motion vector residuals of the picture block being processed.
The third identification information includes indicating that the prediction mode of the picture block to be processed is the merge mode or the skip mode.

Based on the possible implementations described above, different identification information is used to distinguish between different prediction modes, so that the motion vector data of the picture block being processed is stored in multiple prediction modes. Is possible.

In a possible implementation, the step of getting the final predicted motion vector based on the initial predicted motion vector and multiple pre-set offset vectors is
The step of adding the initial motion vector and multiple pre-set offset vectors separately to get multiple candidates for the final motion vector,
It includes a step of determining a final motion vector candidate corresponding to the minimum distortion cost among a plurality of final motion vector candidates as a final motion vector.

Based on the above possible implementation, the distortion cost is used to obtain the final predicted motion vector with higher accuracy, and the stored motion vector data is more accurate.

According to the fourth aspect, the present application provides a motion vector acquisition device configured to execute the above motion vector acquisition method. Specifically, the motion vector acquisition device includes a functional module configured to perform the motion vector acquisition method provided in any one of the first aspect or the optional method of the first aspect. .. The above aspect corresponds to the motion vector acquisition method.

According to a fifth aspect, the present application provides a motion vector residual determination device configured to perform the motion vector residual determination method described above. Specifically, the motion vector residual determination device is configured to perform the motion vector residual determination method provided by any one of the second aspect or the optional method of the second aspect. Includes functional modules. The above aspect corresponds to the motion vector residual determination method.

According to a sixth aspect, the present application provides a motion vector data storage device configured to perform the motion vector data storage method described above. Specifically, the motion vector data storage device is configured to perform the motion vector data storage method provided by any one of the third aspect or the optional method of the third aspect. Includes functional modules that have been created. The above aspect corresponds to the motion vector data storage method.

According to a seventh aspect, the present application provides a computer device. Computer devices include processors and memory. The memory stores at least one instruction, which is loaded and executed by the processor to implement the operations performed in the motion vector acquisition method described above.

According to an eighth aspect, the present application provides a computer device. Computer devices include processors and memory. The memory stores at least one instruction, which is loaded and executed by the processor to perform the operations performed in the motion vector data storage method described above.

According to a ninth aspect, the present application provides a computer device. Computer devices include processors and memory. The memory stores at least one instruction, which is loaded and executed by the processor to implement the operations performed in the motion vector residual determination method described above.

According to a tenth aspect, the present application provides a computer-readable storage medium. The storage medium stores at least one instruction, which is loaded and executed by the processor to realize the operations performed in the motion vector acquisition method described above.

According to the eleventh aspect, the present application provides a computer-readable storage medium. The storage medium stores at least one instruction, which is loaded and executed by the processor to implement the operations performed in the motion vector residual determination method described above.

According to a twelfth aspect, the present application provides a computer-readable storage medium. The storage medium stores at least one instruction, which is loaded and executed by the processor to realize the operations performed in the motion vector data storage method described above.

The technical solutions provided in this application include at least the following beneficial effects:

The initial motion vector of the picture block to be processed is determined by using the positional relationship between the reference block and the picture block to be processed. If the reference block and the picture block to be processed are within a preset range, the decoder uses the initial motion vector of the reference block as the motion vector of the picture block to be processed. If the reference block and the picture block to be processed are out of the preset range, the decoder uses the final motion vector of the reference block as the motion vector of the picture block to be processed. In this case, if the picture block to be processed needs to be decoded by using the final motion vector of the reference block at the time of decoding, the decoder will use the picture block to be processed. The initial motion vector of the reference block can be used as the motion vector of the picture block to be processed so that the block can be used, which gives the final motion vector of the reference block. It is possible to avoid the case where the motion vector of the picture block to be processed can be acquired and improve the decoding efficiency only after the motion vector is obtained.

FIG. 3 is a block diagram of an example video coding system that can be configured for use in embodiments of the present application.

FIG. 3 is a system block diagram of an example video encoder that can be configured for use in embodiments of the present application.

FIG. 3 is a system block diagram of an example video decoder that can be configured for use in embodiments of the present application.

It is a schematic block diagram of an example of an inter-prediction module that can be configured for use in embodiments of the present application.

It is a schematic diagram of an example of a coding unit and adjacent picture blocks associated with the coding unit.

It is a flowchart of an implementation example which constructs a list of predicted motion vectors of candidates.

It is a schematic of the implementation example which adds the combined candidate motion vector to the list of predicted motion vectors of merge mode candidates.

It is a schematic of the implementation example which adds a scaled candidate motion vector to a list of predicted motion vectors of merge mode candidates.

It is a schematic of the implementation example which adds a zero motion vector to a list of predicted motion vectors of merge mode candidates.

It is a flowchart of the implementation example of the merge prediction mode.

It is a flowchart of the implementation example of the advanced motion vector prediction mode.

FIG. 3 is a flow chart of an example motion compensation implementation performed by a video decoder that can be configured for use in embodiments of the present application.

It is a schematic flowchart of the motion vector update method during video coding by embodiment of this application.

It is a schematic flowchart of the motion vector update method during video decoding by embodiment of this application.

It is a schematic flowchart which updates a motion vector by embodiment of this application.

It is a schematic flowchart which updates a motion vector by embodiment of this application.

It is a schematic flowchart which updates a motion vector by embodiment of this application. It is a schematic flowchart which updates a motion vector by embodiment of this application.

It is a schematic flowchart which updates a motion vector by embodiment of this application.

It is a schematic flowchart of the motion vector acquisition method by embodiment of this application.

It is a schematic flowchart of the motion vector acquisition method by embodiment of this application.

It is a schematic diagram which selects the motion vector within the range specified according to the embodiment of this application.

It is a schematic diagram of the acquisition method of the present prediction block by embodiment of this application.

It is a schematic block diagram of the motion vector acquisition apparatus during video decoding by the embodiment of this application.

FIG. 3 is a schematic block diagram of a video coding device according to an embodiment of the present application.

It is a structural drawing of the motion vector acquisition apparatus by embodiment of this application.

It is a structural drawing of the motion vector residual determination apparatus by embodiment of this application.

It is a structural diagram of the motion vector data storage apparatus according to the embodiment of this application.

In order to further clarify the objectives, technical solutions, and advantages of the present application, the implementation of the present application will be described in more detail below with reference to the accompanying drawings.

FIG. 1 is a schematic block diagram of a video coding system 10 according to an embodiment of the present application. As shown in FIG. 1, the system 10 includes a source device 12. The source device 12 produces encoded video data that should later be decoded by the destination device 14. The source device 12 and the destination device 14 are desktop computers, notebook computers, tablet computers, set-top boxes, telephone handset such as "smart" phones, "smart" touchpads, televisions, cameras, display devices. , Digital media players, video game consoles, video streaming transmitters, etc. may include any of a wide range of devices. In some applications, the source device 12 and the destination device 14 may be equipped for wireless communication.

The destination device 14 can receive the encoded video data to be decoded via the link 16. Link 16 may include any type of medium or device capable of transmitting encoded video data from the source device 12 to the destination device 14. In a feasible implementation, the link 16 may include a communication medium that allows the source device 12 to send the encoded video data directly to the destination device 14 in real time. The encoded video data can be modulated according to a communication standard (eg, a wireless communication protocol) and then transmitted to the destination device 14. The communication medium may include any wireless or wired communication medium, such as a radio frequency spectrum or at least one physical transmission line. The communication medium can form part of a packet-based network (eg, a local area network, a wide area network, or a global network such as the Internet). The communication medium may include routers, switches, base stations, or any other device that can be used to facilitate communication from the source device 12 to the destination device 14.

Alternatively, the encoded data may be output from the output interface 22 to the storage device. Similarly, the encoded data may be accessed from the storage device via the input interface. The storage device is a plurality of distributed data storage media or locally accessible data storage media such as hard drives, Blu-rays, digital video discs (DVDs), compact discs read-only. Memory (compact disc read-only memory, CD-ROM), flash memory, volatile or non-volatile memory, or any other suitable digital storage configured to store encoded video data. It may contain any one of the media. In another feasible implementation, the storage device may correspond to a file server or another intermediate storage device capable of holding the encoded video produced by the source device 12. The destination device 14 can access the stored video data from the storage device by streaming transmission or download. The file server can be any type of server capable of storing the encoded video data and transmitting the encoded video data to the destination device 14. In a feasible implementation, file servers include website servers, file transfer protocol servers, network-attached storage devices, or local disk drives. The destination device 14 can access the encoded video data via any standard data connection, including an internet connection. A data connection is a wireless channel (eg, a wireless-fidelity (Wi-Fi) connection) or a wired connection (eg, a wireless-fidelity) suitable for accessing encoded video data stored on a file server. Cable modems), or combinations thereof. The transmission of the encoded video data from the storage device may be a streaming transmission, a download transmission, or a combination thereof.

The techniques in this application are not necessarily limited to wireless applications and settings. The technology is used in multiple multimedia applications such as over-the-air television broadcasting, cable television transmission, satellite television transmission, video streaming transmission (eg, over the Internet), and data storage media. Applicable to video decoding to support any of digital video encoding for storage, decoding of digital video stored on data storage media, or any other application. Is possible. In some feasible implementations, the system 10 will support one-way or two-way video transmission to support applications such as video streaming transmission, video playback, video broadcasting, and / or video telephone. It may be configured in.

In the feasible implementation of FIG. 1, the source device 12 includes a video source 18, a video encoder 20, and an output interface 22. In some applications, the output interface 22 may include a modulator / demodulator (modem) and / or a transmitter. In the source device 12, the video source 18 may include, for example, the following sources: a video capture device (eg, a video camera), a video archive containing previously captured video, and a video receiving video from a video content provider. It can include a feed-in interface and / or a computer graphics system that produces computer graphics data as source video, or a combination thereof. In a feasible implementation, if the video source 18 is a video camera, the source device 12 and the destination device 14 can configure a camera phone or a video phone. For example, the techniques described herein can be applied, for example, to video decoding and can be applied to wireless and / or wired applications.

The video encoder 20 can encode captured or pre-captured video, or computer-generated video. The encoded video data can be transmitted directly to the destination device 14 via the output interface 22 of the source device 12. The encoded video data can also (or alternatively) be stored in the storage device 24 so that the destination device 14 or another device can subsequently be decoded and / or played back. To access the encoded video data.

The destination device 14 includes an input interface 28, a video decoder 30, and a display device 32. In some applications, the input interface 28 may include a receiver and / or a modem. The input interface 28 of the destination device 14 receives the encoded video data via the link 16. The encoded video data transmitted or provided to the storage device 24 over the link 16 is generated by the video encoder 20 and for decoding the video data by a video decoder such as the video decoder 30. May contain multiple syntax elements used in. These syntax elements can be included in the encoded video data transmitted over the communication medium and stored on the storage medium or file server.

The display device 32 may be integrated with the destination device 14 or may be arranged outside the destination device 14. In some feasible implementations, the destination device 14 may include an integrated display device or may be configured to connect to an interface of an external display device. In other feasible implementations, the destination device 14 may be a display device. Generally, the display device 32 displays the decoded video data to the user and is among a plurality of display devices such as a liquid crystal display, a plasma display, an organic light emitting diode display, or another type of display device. Any one can be included.

The video encoder 20 and the video decoder 30 can operate according to, for example, the next generation video coding compression standard (H.266) currently being developed, and H.I. It may be compliant with the 266 test model.

Alternatively, the video encoder 20 and the video decoder 30 are described, for example, by ITU-T H. 265 standard or ITU-T H. It may operate according to other specialized or industrial standards such as the 264 standard, or extensions of these standards. ITU-T H. The 265 standard, also referred to as the high efficiency video decoding standard, is the ITU-T H.D. Alternatively, the 264 standard is also referred to as moving picture experts group (MPEG-4) Part 10 or advanced video coding (AVC). However, the technique of the present application is not limited to any particular decoding standard. In other feasible implementations, video compression standards are MPEG-2 and ITU-TH. 263 is included.

Although not shown in FIG. 1, in some embodiments, the video encoder 20 and the video decoder 30 may be integrated with the audio encoder and audio decoder, respectively, and may be the same data stream or separate. Appropriate multiplexer / demultiplexer (multiplexer-demultiplexer, MUX-DEMUX) units or other hardware and software may be included to encode both audio and video in the data stream. If applicable, in some feasible implementations, the MUX-DEMUX unit is an ITU H. It may comply with the 223 multiplexer protocol, or other protocols such as the user datagram protocol (UDP).

The video encoder 20 and the video decoder 30 include a plurality of suitable encoder circuits, such as one or more microprocessors, digital signal processing (DSP), application specific integrated circuit (ASIC). ), Field programmable gate array (FPGA), discrete logic, software, hardware, firmware, or any combination thereof. be. When some techniques are realized as software, the device stores software instructions in a suitable non-temporary computer-readable medium and uses one or more processors to realize the techniques in this application. By doing so, the instruction can be executed in the form of hardware. The video encoder 20 and the video decoder 30 may be included in one or more encoders or decoders, respectively. Both the video encoder 20 and the video decoder 30 can be integrated as part of a combined encoder / decoder (CODEC) in the corresponding device.

The present application may relate, for example, to another device in which the video encoder 20 "signals" certain information to, for example, the video decoder 30. However, it should be understood that the video encoder 20 can associate a particular syntax element with an encoded portion of the video data in order to signal the information. That is, the video encoder 20 can store a specific syntax element in the header information of the encoded portion of the video data in order to signal the data. In some applications, these syntax elements are encoded and stored (eg, stored in a storage system or file server) before being received and decoded by the video decoder 30. Thus, the term "signal" is used, for example, to transmit or compress a syntax, regardless of whether the transmission takes place in real time, near real time, or within a time span. It may mean the transmission of other data used to decode the video data. For example, transmission can be performed if the syntax element is stored on the medium during coding, after which the syntax element is decoded at any time after it is stored on the medium. It can be taken out by a computer.

The joint video team (JVT) is H.K. 265 High efficiency video coding (HEVC) standard has been developed. The HEVC standardization is based on an evolved model of the video decoder, which is called the HEVC test model. The latest H. The 265 standard document is http: // www. itu. int / rec / T-REC-H. Available at 265. The latest version of the standard document is H. 265 (12/16), the standard document is incorporated herein by reference in its entirety. In HM, the video decoder is ITU-TH. It is hypothesized to have some additional capabilities associated with existing 264 / AVC algorithms. For example, H. The 264 can provide 9 intra-predictive coding modes, while the HM can provide up to 35 intra-predictive coding modes.

JVET is H. We are working on the development of the 266 standard. H. The 266 standardization process is based on an evolved model of the video decoder, which is based on H.D. It is called the 266 test model. H. The explanation of the 266 algorithm is http: // phenix. int−evry. It is available on fr / jvet and the latest algorithm description is contained in JVET-G1001-v1. The entire algorithm description document is incorporated herein by reference in its entirety. In addition, reference software for the JEM test model is available at https: // jvet. hhi. fraunhofer. It is available at de / svn / svn_HMJEM Software / and is incorporated by reference in its entirety into this application.

Generally, as described in the HM working model, a video frame or picture is divided into a series of tree blocks or a largest coding unit (LCU) containing both Luma and Chroma samples. It is possible to be done. The LCU is also called a coding tree unit (CTU). The tree block is H. It has the same function as that of the 264 standard macroblock. The slice contains several contiguous tree blocks in the order of decryption. The video frame or picture can be divided into one or more slices. Each tree block can be divided into coding units based on quadtrees. For example, a tree block that acts as the root node of a quadtree can be split into four child nodes, each child node can also act as a parent node, and the other four. Divided into two child nodes. The final indivisible child node that acts as the leaf node of the quadtree contains a decryption node, eg, a decrypted video block. In the syntax data associated with the decrypted bitstream, the maximum number of times the tree block can be split and the minimum size of the decrypted node may be determined.

It should be noted that one CTU contains one coding tree block (CTB) and two chroma CTBs at the same location and several corresponding syntax elements. The CTB can be used directly as a coding block (CB), or multiple small coding blocks CB, one Luma CB, two Chroma CBs, and several. It is possible to divide into the corresponding syntax elements of the above in the form of a quadtree to form one coding unit (CU).

The CU includes a decoding node, a prediction unit (PU), and a transform unit (TU) associated with the decoding node. The size of the CU corresponds to the size of the decoding node, and the shape of the CU needs to be square. The size of the CU may range from 8x8 pixels up to 64x64 pixels, or may be a larger tree block size. Each CU can include one or more PUs and one or more TUs. For example, CU-related syntax data can describe dividing one CU into one or more PUs. The partitioning pattern can be different if the CU is coded in skip or direct mode, in intra-prediction mode, or in inter-prediction mode. The PU obtained by partitioning may have a non-square shape. For example, CU-related syntax data can also describe the partitioning of a CU to one or more TUs based on a quadtree. The TU may have a square or non-square shape.

The HEVC standard allows TU-based conversion. Different CUs may contain different TUs. The size of the TU is usually set based on the size of the PU in a given CU defined for the partitioned LCU. However, this may not always be the case. The size of the TU is usually the same as or smaller than that of the PU. In some feasible implementations, a quadtree structure called a "residual quadtree" (RQT) is used to divide the CU-corresponding residual sample into smaller units. there is a possibility. Leaf nodes of RQT are sometimes referred to as TUs. Pixel differences associated with the TU may be converted to generate a conversion factor, and the conversion factor may be quantized.

Generally, the PU contains data related to the prediction process. For example, if the PU is encoded in intra mode, the PU may contain data describing the intra prediction mode of the PU. In another feasible implementation, if the PU is encoded in intermode, the PU may contain data defining motion vectors for the PU. For example, the data defining the motion vector for the PU is indicated by the horizontal component of the motion vector, the vertical component of the motion vector, the resolution of the motion vector (eg, 1/4 pixel accuracy or 1/8 pixel accuracy), and the motion vector. A reference picture and / or a motion vector reference picture list (eg, list 0, list 1, or list C) can be described.

Generally, transformation and quantization processes are used for TUs. A given CU containing one or more PUs can also contain one or more TUs. After performing the prediction, the video encoder 20 can calculate the residuals corresponding to the PU. The residuals include pixel differences. Pixel differences can be converted to conversion factors, which are quantized and scanned using the TU to produce a serialized conversion factor for entropy decoding. In the present application, the term "video block" is typically used to indicate a CU decoding node. In some specific applications, the term "video block" can also be used to indicate a tree block containing a decryption node, PU, and TU, such as an LCU or CU.

A video sequence usually includes a series of video frames or pictures. For example, a group of pictures (GOP) includes a series of video pictures, or one or more video pictures. The GOP may include syntax data in the header information of the GOP, the header information of one or more pictures, and the like, and the syntax data describes the amount of pictures contained in the GOP. Each slice of a picture may contain slice syntax data that describes the coding mode of the corresponding picture. The video encoder 20 typically performs an operation on a video block within an individual video slice to encode the video data. The video block may correspond to a decryption node in the CU. The size of the video block may be fixed or variable and may vary depending on the specified decoding standard.

In a feasible implementation, HM supports prediction at various PU sizes. Assuming that the size of a particular CU is 2N × 2N, the HM will make an intra prediction for the PU using a size of 2N × 2N or N × N and 2N × 2N, 2N × N, N × 2N, or N. Supporting inter-prediction for symmetric PUs using xN size, HM also supports inter-prediction asymmetric parties for PUs using xN x nU, 2N x nD, nL x 2N, or nR x 2N sizes. It also supports shining. In asymmetric partitioning, the CU is not divided in one direction, but in two directions, one part occupying 25% of the CU and the other part occupying 75% of the CU. The portion that occupies 25% of the CU is indicated by an indicator that indicates "n" followed by "U (Up)", "D (Down)", "L (Left)", and "R (Right)". Therefore, for example, "2N x nU" refers to a horizontally divided 2N x 2N CU, the upper part is 2N x 0.5N PU, and the lower part is 2N x 1.5N PU.

In the present application, "NxN" and "N times N" indicate the pixel size of a vertical and horizontal dimensional video block, such as 16x16 pixels or the number of pixels obtained by multiplying 16 by 16. It can be used interchangeably for. Generally, a 16x16 block has 16 pixels (y = 16) in the vertical direction and 16 pixels (x = 16) in the horizontal direction. Similarly, an N × N block usually has N pixels in the vertical direction and N pixels in the horizontal direction, where N is a non-negative integer value. The pixels in the block can be arranged in rows and columns. Also, the amount of pixels in the horizontal direction and the amount of pixels in the vertical direction of the block may not always be the same. For example, the block may contain N × M pixels, where M is not necessarily equal to N.

After performing intra-or inter-predictive decoding for the PU in the CU, the video encoder 20 can calculate the residual data of the TU in the CU. The PU may contain pixel data within the spatial domain (also known as the pixel domain), and the TU is a transformation applied to the residual video data (eg, the discrete cosine transform (DCT)). , Integer transform, wavelet transform, or other conceptually similar transform) may include coefficients in the transform domain. The residual data may correspond to the pixel difference between the pixels of the unencoded picture and the predictor corresponding to the PU. The video encoder 20 can generate a TU containing the residual data of the CU and then convert the TU to generate a conversion factor of the CU.

After performing some conversion to generate the conversion factor, the video encoder 20 can quantize the conversion factor. Quantization refers to the process of quantizing the coefficients, for example to reduce the amount of data used to represent the coefficients and to perform further compression. The quantization process can reduce the bit depth associated with some or all of the coefficients. For example, during quantization, the n-bit value can be reduced to an m-bit value by rounding, where n is greater than m.

The JEM model further improves the video-picture coding structure. Specifically, a block coding structure called a "quad tree combined with binary tree (QTBT) structure" will be introduced. Without using concepts like CU, PU, TU in HEVC, the QTBT structure supports a more flexible CU split shape. The CU may be square or rectangular. Quadtree partitioning is first performed on the CTU, and binary partitioning is further performed on the leaf node of the quadtree. In addition, there are two binary partitioning modes: symmetric horizontal partitioning and symmetric vertical partitioning. The binary tree leaf node is referred to as the CU. The CU in the JEM model cannot be further subdivided between prediction and transformation. In other words, the CU, PU, and TU in the JEM model have the same block size. In the existing JEM model, the maximum CTU size is 256 x 256 Luma pixels.

In some feasible implementations, the video encoder 20 scans the quantized conversion factors in a predetermined scan order to produce a serialized vector that can be entropy-coded. Can be done. In other feasible implementations, the video encoder 20 can perform adaptive scanning. After scanning the quantized transformation coefficients to form a one-dimensional vector, the video encoder 20 is subjected to context-adaptive variable-length coding (CAVLC) method, context-based adaptive binary. Entropy decoding a one-dimensional vector based on the context-based adaptive binary arithmetic coding (CABAC) method, the probability interval partitioning entropy (PIPE) coding method, or another entropy coding method. can do. The video encoder 20 can further encode the syntax elements associated with the encoded video data, so that the video decoder 30 decodes the video data.

To perform CABAC, the video encoder 20 can assign a context in the context model to a symbol to be transmitted. The context can be associated with whether the adjacent value of the symbol is non-zero. To perform CAVLC, the video encoder 20 can select the variable length code of the symbol to be transmitted. Codewords (variable-length coding, VLC) in variable length coding are configured such that shorter codes correspond to more likely symbols and longer codes correspond to less likely symbols. Is possible. In this way, using VLC can reduce the bit rate as compared to using codewords of equal length for all transmitted symbols. Probabilities in CABAC can be determined based on the context assigned to the symbol.

In this embodiment of the present application, the video encoder can perform inter-prediction to reduce temporal redundancy between pictures. As mentioned above, the CU can have one or more prediction unit PUs according to different video compression coding standards. In other words, multiple PUs may belong to one CU, or the PUs and CUs have the same size. As used herein, if the CU and PU have the same size, the partitioning mode of the CU either performs no partitioning or divides the CU into one PU, where the PUs are uniform for illustration purposes. Used for. When the video encoder performs inter-prediction, the video encoder can signal motion information about the PU to the video decoder. For example, motion information about a PU may include a reference picture index, a motion vector, and a predictive direction identifier. The motion vector can indicate the displacement between the picture block of the PU (also called a video block, pixel block, pixel set, etc.) and the reference block of the PU. The reference block of the PU may be part of a reference picture similar to the picture block of the PU. The reference block may be placed within the reference picture indicated by the reference picture index and the predictive direction identifier.

To reduce the amount of coded bits needed to represent motion information about the PU, the video encoder follows a merge prediction mode or advanced motion vector prediction (AMVP) mode. A list of candidate predicted motion vectors (MVs) for each PU can be generated. Each of the candidate predicted motion vectors in the list of candidate predicted motion vectors for the PU can indicate motion information. The motion information represented by some candidate predicted motion vectors in the list of candidate predicted motion vectors may be based on motion information for other PUs. Candidate prediction if the candidate's predicted motion vector indicates motion information for one of a particular spatial candidate's predicted motion vector position or a particular temporal candidate's predicted motion vector position. The resulting motion vector may be referred to herein as the predicted motion vector of the "original" candidate. For example, in the merge mode, which is also referred to as the merge prediction mode in the present application, there may be a predicted motion vector position of five original space candidates and a predicted motion vector position of one original time candidate. In some examples, the video encoder may combine several motion vectors from different original candidate predicted motion vectors, modify the original candidate's predicted motion vector, or predict the candidate. By inserting only the zero motion vector as the motion vector, the predicted motion vector of additional candidates can be generated. The predicted motion vectors of these additional candidates are not considered to be the predicted motion vectors of the original candidate and may be referred to herein as the manually generated predicted motion vectors of the candidates.

Techniques in the present application typically include techniques for generating a list of predicted motion vectors for candidates in a video encoder and techniques for generating a list of predicted motion vectors for the same candidate in a video decoder. Video encoders and video decoders can generate a list of predicted motion vectors for the same candidate by implementing the same technique for constructing a list of predicted motion vectors for the same candidate. For example, a video encoder and a video decoder can construct a list with the same amount of candidate predicted motion vectors (eg, 5 candidate predicted motion vectors). Video encoders and video decoders first consider the predicted motion vectors of spatial candidates (eg, adjacent blocks in the same picture), and then (eg, predicted candidates in different pictures). Considering the predicted motion vectors of temporal candidates (such as motion vectors) and finally considering the predicted motion vectors of manually generated candidates, the required amount of candidate predictions. This can be done until the motion vector is added to the list. According to the technique of the present application, there is to remove the iterative candidate's predicted motion vector from the candidate's list of predicted motion vectors while constructing the candidate's list of predicted motion vectors. A pruning process may be performed on the predicted motion vector of the type candidate, and the pruning process is performed on the predicted motion vector of the other type candidate in order to reduce the complexity of the decoder. It does not have to be. For example, for a set of predicted motion vectors of spatial candidates and a set of predicted motion vectors of temporal candidates, the pruning process combines the predicted motion vectors of the candidates with iterative motion information. It can be performed to remove from the list of candidate predicted motion vectors. However, the manually generated candidate predicted motion vector is added to the list of candidate predicted motion vectors if the pruning process is not performed on the manually generated candidate predicted motion vector. It is possible.

After generating a list of candidate predicted motion vectors for the PU of the CU, the video encoder selects the candidate predicted motion vector from the list of candidate predicted motion vectors and the candidate predicted motion vector. The motion vector index can be output as a bit stream. The predicted motion vector of the selected candidate may be a candidate predicted motion vector for generating a motion vector that best matches the predicted unit of the decoded target PU. The candidate's predicted motion vector index can indicate the position of the selected candidate's predicted motion vector in the list of candidate's predicted motion vectors. The video encoder may further generate a predictive picture block for the PU based on the reference block indicated by the motion information about the PU. The motion information about the PU may be determined based on the motion information indicated by the predicted motion vector of the selected candidate. For example, in merge mode, the motion information of the PU may be the same as the motion information indicated by the predicted motion vector of the selected candidate. In the advanced motion vector prediction mode, the motion information for the PU can be determined based on the motion vector difference for the PU and the motion information indicated by the predicted motion vector of the selected candidate. .. The video encoder can generate one or more residual picture blocks for the CU based on the predicted picture blocks for the PU of the CU and the original picture blocks for the CU. The video encoder can then encode one or more residual picture blocks and output one or more residual picture blocks in a bitstream.

The bitstream can contain data that identifies the predicted motion vectors of the selected candidates in the list of predicted motion vectors of the candidates for the PU. The video decoder can determine the motion information of the PU based on the motion information indicated by the predicted motion vector of the selected candidate in the candidate predicted motion vector list for the PU. The video decoder can identify one or more reference blocks for the PU based on the motion information of the PU. After identifying one or more reference blocks for the PU, the video decoder may generate a predictive picture block for the PU based on the one or more reference blocks for the PU. The video decoder can reconstruct the picture block for the CU based on the predicted picture block for the PU of the CU and one or more residual picture blocks for the CU.

For the sake of brevity, the location or picture block can be described herein as having various spatial relationships with the CU or PU. The description can be described as follows: The location or picture block has various spatial relationships with the picture block associated with the CU or PU. Additionally, in the present application, the PU currently being decoded by the video decoder may be referred to as the PU to be processed, or may be referred to as the picture block currently being processed. In the present application, the CU currently decoded by the video decoder may be referred to as the CU to be processed. In the present application, the picture currently decoded by the video decoder may be referred to as the current picture. It should be understood that the present application is also applicable if the PU and CU have the same size, or if the PU is a CU. PU is used uniformly for illustration purposes.

As briefly described above, the video encoder 20 can generate predictive picture blocks and motion information for the PU of the CU by inter-prediction. In many examples, the motion information of a given PU is of one or more adjacent PUs (ie, a PU in which a picture block is spatially or temporally adjacent to a picture block of a given PU). It may be the same as or similar to the motion information. Since the adjacent PUs often have similar motion information, the video encoder 20 can encode the motion information of a given PU based on the motion information of the adjacent PUs. Encoding the motion information of a given PU based on the motion information of adjacent PUs reduces the amount of encoded bits required in the bitstream to indicate the motion information of a given PU. Can be done.

The video encoder 20 can encode the motion information of a given PU based on the motion information of the adjacent PUs in various ways. For example, the video encoder 20 can indicate that the motion information for a given PU is the same as the motion information for an adjacent PU. In the present application, the merge mode may be used to indicate that the motion information for a given PU is the same as the motion information for an adjacent PU, or it may be derived from the motion information for an adjacent PU. good. In another feasible implementation, the video encoder 20 is capable of calculating a motion vector difference (MVD) for a given PU, and the MVD is an initial motion vector for a given PU. And the difference between the final motion vector for a given PU. The video encoder 20 can include the MVD in place of the motion vector of a given PU in the motion information of a given PU. In a bitstream, the amount of coded bits needed to represent an MVD is less than the amount of coded bits needed to represent a motion vector for a given PU. The present application uses an advanced motion vector prediction mode to show that motion information for a given PU is signaled to the decoder by using an MVD and an index value to identify a candidate motion vector. be able to.

In order to signal the motion information of a given PU to the decoder in merge mode or AMVP mode, the video encoder 20 can generate a list of candidate predicted motion vectors for a given PU. The list of candidate predicted motion vectors may include one or more candidate predicted motion vectors. Each candidate's predicted motion vector in the list of candidate predicted motion vectors for a given PU can specify motion information. The motion information represented by each candidate's predicted motion vector may include a motion vector, a reference picture index, and a predicted direction identifier. The predicted motion vectors of the candidates in the list of predicted motion vectors of the candidates may contain the predicted motion vectors of the "original" candidates, and the predicted motion vectors of each "original" candidate , Shows motion information of one of the predicted motion vector positions of a particular candidate in a PU different from a given PU.

After generating a list of candidate predicted motion vectors for the PU, the video encoder 20 picks one of the candidate predicted motion vectors from the list of candidate predicted motion vectors for the PU. You can choose. For example, the video encoder can compare each candidate's predicted motion vector to the decoded PU and select a candidate's predicted motion vector with the desired rate distortion cost. The video encoder 20 can output a candidate predicted motion vector index for the PU. The candidate's predicted motion vector index can identify the position of the selected candidate's predicted motion vector in the candidate's predicted motion vector list.

Further, the video encoder 20 may generate a predicted picture block of the PU based on the reference block indicated by the motion information of the PU. The motion information for the PU may be determined based on the motion information indicated by the predicted motion vector of the candidate selected in the list of predicted motion vectors for the candidate for the PU. For example, in merge mode, the motion information of the PU may be the same as the motion information indicated by the predicted motion vector of the selected candidate. In the AMVP mode, the motion information of the PU may be determined based on the motion vector difference of the PU and the motion information indicated by the predicted motion vector of the selected candidate. As mentioned above, the video encoder 20 can process the predicted picture block of the PU.

When the video decoder 30 receives the bitstream, the video decoder 30 can generate a list of candidate predicted motion vectors for each PU of the CU. The list of candidate predicted motion vectors generated by the video decoder 30 for the PU may be the same as the list of candidate predicted motion vectors generated by the video encoder 20 for the PU. The syntax element obtained by analyzing the bitstream can indicate the position of the predicted motion vector of the selected candidate in the list of predicted motion vectors of the candidate for the PU. After generating a list of candidate predicted motion vectors for the PU, the video decoder 30 may also generate a PU predictive picture block based on one or more reference blocks indicated by the PU motion information. good. The video decoder 30 can determine the motion information of the PU based on the motion information indicated by the predicted motion vector of the candidate selected in the list of predicted motion vectors of the candidates for the PU. The video decoder 30 can reconstruct the picture block of the CU based on the predicted picture block of the PU and the residual picture block of the CU.

In a feasible implementation, the bits for constructing a list of candidate predicted motion vectors and obtaining the position of the selected candidate's predicted motion vector in the candidate's list of predicted motion vectors on the decoder. It should be understood that the analysis of the stream is independent of each other and may be performed in any order or in parallel.

In another feasible implementation, on the decoder, the position of the predicted motion vector of the candidate selected in the list of predicted motion vectors of the candidate is first obtained by analyzing the bit stream and then the candidate. A list of predicted motion vectors of is constructed based on the positions obtained by the analysis. In this implementation, it is not necessary to build a list of predicted motion vectors for all candidates, only the list of predicted motion vectors for candidates at the position obtained by the analysis is the candidate at that position. It needs to be constructed to determine the predicted motion vector. For example, by analyzing the bit stream, the predicted motion vector of the selected candidate is the predicted motion vector of the candidate having an index of 3 in the list of predicted motion vectors of the candidate and having an index of 3. It can be seen that in order to determine the predicted motion vector of the candidate, it is necessary to construct only the list of the predicted motion vector of the candidate from index 0 to index 3. This can reduce complexity and improve decoding efficiency.

FIG. 2 is a schematic block diagram of the video encoder 20 according to the embodiment of the present application. The video encoder 20 can perform intra-decoding and inter-decoding on the video blocks in the video slice. Intra-decoding relies on spatial prediction to reduce or eliminate spatial redundancy of video within a given video frame or picture. Interdecoding relies on temporal prediction to reduce or eliminate temporal redundancy of video in adjacent frames or pictures of a video sequence. The intra mode (I mode) may be any one of several spatial compression modes. The inter-mode, such as one-way prediction mode (P mode) or two-way prediction mode (B mode), may be any one of several temporal compression modes.

In a feasible implementation of FIG. 2, the video encoder 20 includes a partitioning unit 35, a prediction unit 41, a reference picture storage 64, an adder 50, a conversion unit 52, a quantization unit 54, and an entropy coding unit 56. include.

The reference picture storage 64 provides a target storage space for storing the MVD, i.e., to store the difference between the final motion vector and the initial motion vector for the PU to be processed. , Further configured. In some embodiments, the reference picture storage 64 can further store the final motion vector of the PU to be processed. In this embodiment of the present application, the MVD stored in the reference picture storage 64 may be used in another PU coding process. For example, in the decryption process in the merge mode, the unupdated initial motion vector and the updated final motion vector are stored separately at different times. Therefore, if the PU to be processed needs to obtain the predicted motion vector for another PU, under certain conditions, the PU to be processed will get the unupdated initial motion vector of the other PU. , It can be used directly as a motion vector of another PU, and it is not necessary to obtain the motion vector of another block after the motion vector of another PU is updated.

Note that the target storage space may be provided by memory other than the reference picture storage 64, eg, newly added memory. This is not specifically limited in this embodiment of the present application.

In a related technique, when the video encoder 20 operates in merge mode, the contents of the reference picture storage 64 are empty or zero vector, i.e., the video encoder 20 is in merge mode. When operating, the target storage space is not used during encoding, and in the encoding process, the predicted motion vector of the other PUs obtained by the PU being processed is the updated final of the other PUs. Note that it is a motion vector. Therefore, the PU to be processed can obtain the predicted motion vector for the other PU only after the predicted motion vector of the other PU has been updated, and as a result, the coding during the coding. There is a delay in conversion. However, if the video encoder 20 operates in non-merge mode, the reference picture storage 64 may be used. For example, in AMVP mode, the target storage space is used to store the motion vector residuals, and the video encoder 20 encodes the motion vector residuals and, as a result, after decoding the current PU. The video decoder 30 acquires the motion vector residual of the PU to be processed, and is the final of the PU to be processed based on the motion vector residual of the PU to be processed and the initial motion vector of the PU to be processed. Motion vector can be obtained.

The prediction unit 41 includes a motion estimation unit 42, a motion compensation unit 44, and an intra prediction unit 46. For video block reconstruction, the video encoder 20 further includes an inverse quantization unit 58, an inverse transformation unit 60, and an adder 62. The video encoder 20 may further include a deblocking filter (not shown in FIG. 2) to perform filtering on the block boundaries and remove blocking artifacts from the reconstructed video. If necessary, the deblocking filter typically performs filtering on the output of adder 62. In addition to the deblocking filter, additional loop filters (inside or after the loop) may be used.

As shown in FIG. 2, the video encoder 20 receives the video data and the partitioning unit 35 divides the data into video blocks. Such partitioning may further include partitioning into slices, picture blocks, or other larger units and video block partitioning based on the LCU and CU quadtree structure. For example, the video encoder 20 is a component for encoding a video block in a video slice to be encoded. Usually, one slice can be divided into multiple video blocks (and can be divided into a set of video blocks called picture blocks).

The prediction unit 41 may have a plurality of possibilities for the current video block based on the coding quality and the cost calculation result (eg, rate distortion cost, RDcost). Decoding mode, for example, one of a plurality of intra-decoding modes or one of a plurality of inter-decoding modes can be selected. The prediction unit 41 provides the acquired intra-decoding block or inter-decoding block to the adder 50 to generate residual block data, and the obtained intra-decoding block or inter-decoding block is added to the adder 62. The encoded block can be reconstructed and the reconstructed encoded block can be used as a reference picture.

The motion estimation unit 42 and motion compensation unit 44 in the prediction unit 41 perform interpredictive decoding for the current video block for one or more predictive blocks of one or more reference pictures to perform temporal compression. .. The motion estimation unit 42 may be configured to determine the inter-prediction mode of the video slice based on the preset mode of the video sequence. In preset mode, the video slices in the sequence can be designated as P slices, B slices, GPB slices. The motion estimation unit 42 and the motion compensation unit 44 may be highly integrated, but are described separately to illustrate the concept. The motion estimation performed by the motion estimation unit 42 is a process of generating a motion vector for estimating a video block. For example, the motion vector can indicate the displacement of the PU of the current video frame or video block of the picture with respect to the predicted block of the reference picture.

A predictive block is a block of PUs that are found to be closely matched to the video block to be decoded based on the pixel difference, and the pixel difference is the sum of absolute difference (SAD). ), Sum of squared difference (SSD), or other difference metric can be used. In some feasible implementations, the video encoder 20 can calculate the value of the sub-integer pixel position of the reference picture stored in the reference picture storage 64. For example, the video encoder 20 can interpolate a value at a 1/4 pixel position, a 1/8 pixel position, or another fractional pixel position of a reference picture. Therefore, the motion estimation unit 42 can execute a motion search for all pixel positions and fractional pixel positions and output a motion vector with fractional pixel accuracy.

The motion estimation unit 42 calculates the motion vector for the PU of the video block in the interdecoded slice by comparing the position of the PU with the position of the predicted block of the reference picture. The reference picture can be selected from the first reference picture list (list 0) or the second reference picture list (list 1). Each list is used to identify one or more reference pictures stored in the reference picture storage 64. The motion estimation unit 42 sends the calculated motion vector to the entropy coding unit 56 and the motion compensation unit 44.

Motion compensation performed by motion compensation unit 44 can include extraction or generation of predictive blocks based on motion vectors determined by motion estimation and can perform interpolation with sub-pixel level accuracy. It is possible. After receiving the motion vector for the PU of the current video block, the motion compensation unit 44 can locate the predictive block that the motion vector points to in one reference picture list. The video encoder 20 subtracts the pixel value of the predicted block from the pixel value of the current video block being decoded to obtain the residual video block and obtains the pixel difference. The pixel difference constitutes the residual data of the block and may contain both the Luma difference component and the Chroma difference component. The adder 50 is one or more components that perform a subtraction operation. The motion compensation unit 44 can further generate the video blocks and the syntax elements associated with the video slices, so that the video decoder 30 decodes the video blocks in the video slices.

When the PU is located in the B slice, the picture containing the PU can be associated with two reference picture lists referred to as "list 0" and "list 1". In some feasible implementations, a picture containing a B slice may be associated with a combination of lists in Listing 0 and Listing 1.

Further, when the PU is located in the B slice, the motion estimation unit 42 can execute one-way prediction or two-way prediction for the PU. In some feasible implementations, bidirectional prediction is a prediction performed separately based on the pictures in reference picture list 0 and reference picture list 1. In some other feasible implementation, bidirectional prediction is a prediction that is performed separately based on the reconstructed future frame and the reconstructed past frame for the current frame in display order. be. Further, when the motion estimation unit 42 executes one-way prediction for the PU, the motion estimation unit 42 searches the reference picture in the list 0 or the list 1 in order to search for the reference block for the PU. Can be done. Then, the motion estimation unit 42 can generate a reference index indicating a reference picture including a reference block in list 0 or list 1, and a motion vector indicating a spatial displacement between the PU and the reference block. The motion estimation unit 42 can output a reference index, a prediction direction identifier, and a motion vector as motion information of the PU. The predictive direction identifier may indicate that the reference index indicates a reference picture in Listing 0 or Listing 1. The motion compensation unit 44 can generate a predicted picture block of the PU based on the reference block indicated by the motion information of the PU.

When the motion estimation unit 42 makes bidirectional predictions for the PU, the motion estimation unit 42 searches for a reference picture in list 0 for a reference block for the PU, and further for another reference block for the PU. The reference picture in the list 1 can be searched. Then, the motion estimation unit 42 can generate a reference index indicating a reference picture including the reference blocks of List 0 and Listing 1, and a motion vector indicating a spatial displacement between the reference block and the PU. Further, the motion estimation unit 42 can output the reference index and the motion vector of the PU as the motion information of the PU. The motion compensation unit 44 can generate a predicted picture block of the PU based on the reference block indicated by the motion information of the PU.

In some feasible implementations, the motion estimation unit 42 does not output a complete set of PU motion information to the entropy coding module 56. Instead, the motion estimation unit 42 may signal PU motion information regarding motion information of another PU. For example, the motion estimation unit 42 can determine that the motion information of the PU is similar to the motion information of the adjacent PU. In this implementation, the motion estimation unit 42 can indicate an indicator value in the syntax structure related to the PU, and the indicator value is that the motion information of the PU is the same as the motion information of the adjacent PU. , Or the video decoder 30 may be derived from motion information of adjacent PUs. In another implementation, the motion estimation unit 42 can discriminate between the predicted motion vector of a candidate associated with an adjacent PU and the motion vector difference (MVD) in the PU-related syntax structure. can. The MVD indicates the difference between the motion vector of the PU and the predicted motion vector of the indicated candidate associated with the adjacent PU. The video decoder 30 can determine the motion vector of the PU by using the predicted motion vector and MVD of the indicated candidate.

As mentioned above, the prediction module 41 can generate a list of candidate predicted motion vectors for each PU of the CU. The list of predicted motion vectors for one or more candidates is one or more additions derived from the predicted motion vectors for one or more original candidates and the predicted motion vectors for one or more original candidates. Can include candidate predicted motion vectors.

In this embodiment of the present application, when setting a list of candidate predicted motion vectors in merge mode, the prediction module 41 adds the candidate predicted motion vector list to the reference PU of the PU to be processed. The motion vector can be stored and the motion vector of the reference PU may be the unupdated initial motion vector of the reference PU or the updated final motion vector of the reference PU. It should be noted that it is good. The updated final motion vector for the reference PU may be obtained based on the motion vector stored in the target storage space. If the target storage space stores the MVD, the final motion vector of the reference PU may be obtained by adding the initial motion vector of the reference PU to the MVD stored in the target storage space. It is possible. If the target storage space stores the final motion vector for the reference PU, the final motion vector for the reference PU can be obtained directly from the target storage space. The motion vector of the reference PU can be stored in the list of candidate predicted motion vectors in the following way: if the reference PU and the current PU are in the same CTB or CTB row range. Is stored in the list of candidate predicted motion vectors for the initial unupdated motion vector of the reference unit, or if the reference PU and the current PU are in the same CTB or CTB row range. Stores the updated final motion vector of the reference unit in the list of candidate predicted motion vectors.

The intra prediction unit 46 of the prediction unit 41 performs intra prediction decoding for the current video block for one or more adjacent blocks in the same picture or slice as the current block to be decoded. Can be performed and provide spatial compression. Therefore, as an alternative to the inter-prediction performed by the motion estimation unit 42 and the motion compensation unit 44 (as described above), the intra-prediction unit 46 can perform the intra-prediction on the current block. Specifically, the intra prediction unit 46 can determine the intra prediction mode for encoding the current block. In some feasible implementations, the intra-prediction unit 46 (eg) uses various intra-prediction modes to encode the current block during separate coding traversals, and the intra-prediction unit 46 (or). The mode selection unit 40) in some feasible implementations can select the appropriate intra-prediction mode from the test mode.

After the prediction unit 41 generates a prediction block of the current video block by inter-prediction or intra-prediction, the video encoder 20 subtracts the prediction block from the current video block to obtain the residual video block. do. The residual video data in the residual block is contained in one or more TUs and can be applied to the conversion unit 52. The transform unit 52 performs a transform such as a discrete cosine transform (DCT) or a conceptually similar transform (eg, a discrete sine transform (DST)) to perform a residual video. Convert the data to the residual transform coefficient. The conversion unit 52 can convert the residual video data from the pixel domain to the conversion domain (eg, the frequency domain).

The conversion unit 52 can send the acquired conversion coefficient to the quantization unit 54. The quantization unit 54 quantizes the conversion factor to further reduce the bit rate. The quantization process can reduce the bit depth associated with some or all of the coefficients. The degree of quantization can be modified by adjusting the quantization parameters. In some feasible implementations, the quantized unit 54 can then scan the matrix containing the quantized transformation factors. Alternatively, the entropy coding unit 56 may perform the scan.

After quantization, the entropy coding unit 56 can entropy code the quantized conversion factors. For example, the entropy coding unit 56 may include context-adaptive variable-length coding (CAVLC), context-adaptive binary arithmetic coding (CABAC), probability interval partitioning entropy (PIPE) coding, or other entropy coding methods or techniques. Can be executed. The entropy coding unit 56 can further entropy code the motion vector of the current video slice being decoded and another syntax element. After being entropy-coded by the entropy coding unit 56, the encoded bitstream may be transmitted to the video decoder 30 or recorded for subsequent transmission or retrieval by the video decoder 30. May be good.

The entropy coding unit 56 can encode information indicating the selected intra prediction mode according to the technique of the present application. The video encoder 20 is also referred to as a plurality of intra-prediction mode index tables and a plurality of modified intra-prediction mode index tables (also referred to as codeword mapping tables) used for each of the contexts. ) Can be included in the transmitted bitstream configuration data, such as definitions of coding contexts for various blocks, indications for most probable mode (MPM), intra-predictive mode index tables, and It can contain a modified intra-prediction mode index table.

The inverse quantization unit 58 and the inverse transformation unit 60 perform inverse quantization and inverse transformation, respectively, to reconstruct the residual block in the pixel domain, which is subsequently used as the reference block of the reference picture. The motion compensation unit 44 can calculate the reference block by adding the residual block and the prediction block of one reference picture in one reference picture list. Further, the motion compensation unit 44 may apply one or more interpolation filters to the reconstructed residual block to calculate a sub-integer pixel value for motion estimation. The adder 62 adds the reconstructed residual block and the motion compensated prediction block generated by the motion compensation unit 44 to generate a reference block stored in the reference picture storage 64. The reference block may be used by the motion estimation unit 42 and the motion compensation unit 44 as a reference block for performing interprediction with respect to a block in a subsequent video frame or picture.

FIG. 3 is a schematic block diagram of the video decoder 30 according to the embodiment of the present application. In a feasible implementation of FIG. 3, the video decoder 30 includes an entropy decoding unit 80, a prediction unit 81, an inverse quantization unit 86, an inverse conversion unit 88, an adder 90, and a reference picture storage 92.

The reference picture storage 92 can be configured to provide a target storage space. The target storage space is used to store the MVD, i.e., to store the difference between the final motion vector and the initial motion vector. In some embodiments, the target storage space is used to store the final motion vector. In this embodiment of the present application, the updated final motion vector, or MVD stored in the target storage space, can be used in another PU coding process. For example, in the decryption process in merge mode, the unupdated initial motion vector and the updated final motion vector are stored separately at different times. Therefore, if the PU to be processed needs to obtain the predicted motion vector for another PU, under certain conditions, the PU to be processed will get the unupdated initial motion vector of the other PU. , It can be used directly as a motion vector of another PU, and it is not necessary to obtain the motion vector of another block after the motion vector of another PU is updated.

It should be noted that the target storage space may be provided by memory other than the reference picture storage 92, eg, newly added memory. This is not specifically limited to the embodiments of the present application.

In a related technique, when the video decoder 30 operates in merge mode, the contents of the target storage space are empty or zero vector, that is, the video encoder 20 operates in merge mode. If so, the target storage space is not used during decoding, and during decoding, the predicted motion vectors of the other PUs obtained by the PU being processed are updated by the other PUs. It should be noted that it is the final motion vector. Therefore, the PU to be processed can obtain the predicted motion vector for the other PU only after the predicted motion vector for the other PU has been updated, and as a result, decode during decoding. There will be a delay in conversion. However, if the video decoder 30 operates in non-merge mode, the target storage space may be used. For example, in AMVP mode, the video decoder 30 can obtain the MVD for the PU to be processed by decoding, so that the video decoder 30 can obtain the final motion vector and the MVD for the PU to be processed. Based on, the final motion vector for the PU to be processed can be obtained.

The prediction unit 81 includes a motion compensation unit 82 and an intra prediction unit 84. In some feasible implementations, the video decoder 30 can perform an exemplary decoding process that is the reverse of the coding process described for the video encoder 20 of FIG.

During decoding, the video decoder 30 receives from the video encoder 20 an encoded video bitstream representing a video block of encoded video slices and associated syntax elements. The entropy decoding unit 80 of the video decoder 30 decodes the bitstream to generate quantized coefficients, motion vectors, and other syntax elements. The entropy decoding unit 80 transfers the motion vector and other syntax elements to the prediction unit 81. The video decoder 30 can receive syntax elements at the video slice level and / or the video block level.

If the video slice is decoded into an intra-decrypted (I) slice, the intra-prediction unit 84 of the prediction unit 81 is decoded prior to the signaled intra-prediction mode and the current frame or picture. Predictive data for the video block of the current video slice can be generated based on the data of the block.

When the video picture is decoded into an inter-decoded slice (eg, a B-slice, a P-slice, or a GPB slice), the motion compensation unit 82 of the prediction unit 81 receives the motion received from the entropy-decoding unit 80. Generates a predictive block of the video block of the current video picture based on vectors and other syntax elements. The prediction block may be generated from one reference picture in one reference picture list. The video decoder 30 can use the default configuration technique to construct the reference picture list (List 0 and List 1) based on the reference picture stored in the reference picture storage 92.

The motion compensation unit 82 determines the prediction information of the video block of the current video slice by analyzing the motion vector and other syntax elements, and uses the prediction information to decode the video block. Generate a prediction block for. For example, motion compensation unit 82 uses some of the received syntax elements to decode a video block of a video slice in a predictive (eg, intra-predictive or inter-predictive) mode, inter-predictive slice type. (For example, B slice, P slice, or GPB slice) Configuration information of one or more reference picture lists of the slice, motion vector of each inter-encoded video block of the slice, inter-decoding of each of the slices. Determines the interpredicted status of the video block that has been made, as well as other information for decoding the video block in the current video slice.

The motion compensation unit 82 may further perform interpolation using an interpolation filter. The motion compensation unit 82 can calculate the interpolated value of the sub-integer pixels of the reference block, for example, using the interpolation filter used by the video encoder 20 during the coding of the video block. In the present application, the motion compensation unit 82 can determine the interpolation filter used by the video encoder 20 based on the received syntax element, and use the interpolation filter to generate a prediction block.

If the PU is encoded by inter-prediction, the motion compensation unit 82 can generate a list of candidate predicted motion vectors for the PU. The bitstream may contain data for identifying the position of the predicted motion vector of the selected candidate in the list of predicted motion vectors of the candidate for the PU. After generating the predicted motion vector of the candidate for the PU, the motion compensation unit 82 can generate the predicted picture block for the PU based on one or more reference blocks indicated by the motion information for the PU. The reference block of the PU may be located in a temporal picture different from the temporal picture of the PU. The motion compensation unit 82 can determine the motion information of the PU based on the motion information selected in the list of predicted motion vectors of the candidates for the PU.

In an embodiment of the present application, the motion compensation unit 82 predicts a candidate motion vector with respect to a reference PU of the PU to be processed when setting a list of candidate predicted motion vectors in merge mode. The motion vector may be stored in the list of motion vectors, and the predicted motion vector of the reference PU may be the initial motion vector of the reference PU that has not been updated, or the predicted motion of the reference PU may be updated. It should be noted that it may be a vector. The updated final motion vector of the reference PU may be obtained based on the motion vector for the reference PU stored in the target storage space. For example, if the target storage space stores the MVD, the final motion vector for the reference PU is by adding the initial motion vector for the reference PU and the MVD stored in the target storage space. , Can be obtained. If the target storage space stores the final motion vector for the reference PU, the final motion vector for the reference PU can be obtained directly from the target storage space. .. The predicted motion vector for the reference PU may be stored in the list of candidate predicted motion vectors in the following way: the reference PU and the current PU are in the same CTB or CTB row range. In some cases, the unupdated initial motion vector for the reference unit is stored in the list of candidate predicted motion vectors; or if the reference PU and the current PU are in the same CTB or CTB row range. , The updated final motion vector for the reference unit is stored in the list of candidate predicted motion vectors.

The dequantization unit 86 performs dequantization (eg, dequantization) on the quantization conversion coefficients provided in the bitstream and decoded by the entropy decoding unit 80. The dequantization process determines the degree of quantization based on the quantization parameters calculated by the video encoder 20 for each video block in the video slice, as well as the degree of dequantization to be applied. It is possible to include that. The inverse transformation unit 88 performs an inverse transformation on the transformation coefficients (eg, an inverse DCT, an inverse integer transformation, or a geometrically similar inverse transformation process) to generate a residual block in the pixel domain.

After the motion compensation unit 82 has generated a predictive block of the current video block based on the motion vector and other syntax elements, the video decoder 30 has a residual block from the inverse conversion unit 88 and motion compensation. Performs an addition operation with the corresponding predictive block generated by the unit 82 to form the decoded video block. The adder 90 is one or more components that perform an addition operation. If desired, a deblocking filter may be used to perform filtering on the decoded blocks to remove blocking artifacts. Another loop filter (inside or after the decoding loop) may be used to smooth the pixels, or the video quality may be improved in another way. The decoded video block within a given frame or picture is then stored in reference picture storage 92. The reference picture storage 92 stores a reference picture used for subsequent motion compensation. The reference picture storage 92 also stores the decoded video that is subsequently presented in a display device such as the display device 32 of FIG.

As mentioned above, the techniques of the present application relate, for example, to inter-decoding. The techniques of the present application may be performed by any video decoder described herein, the video decoders being (eg) the video encoder 20 and the video decoder 30 shown and described in FIGS. 1 to 3. It should be understood to include. Specifically, in a feasible implementation, the prediction unit 41 described in FIG. 2 implements the specific techniques described below when inter-prediction is performed while encoding a block of video data. Can be executed. In another feasible implementation, the prediction unit 81 described in FIG. 3 performs the specific technique described below when inter-prediction is performed during decoding of a block of video data. Can be done. Thus, a reference to a general "video encoder" or "video decoder" may include a video encoder 20, a video decoder 30, or another video coding or coding unit.

FIG. 4 is a schematic block diagram of an inter-prediction module according to an embodiment of the present application. The inter-prediction module 121 can include, for example, a motion estimation unit 42 and a motion compensation unit 44. The relationship between PU and CU depends on the video compression coding standard. The inter-prediction module 121 can divide the current CU into PUs according to a plurality of partitioning patterns. For example, the inter-prediction module 121 can classify the current CU into PUs according to a 2Nx2N, 2NxN, Nx2N, NxN partitioning pattern. In another embodiment, the current CU is the current PU. Not limited to this.

The inter-prediction module 121 can perform integer motion estimation (IME) and then fractional motion estimation (FME) for each PU. When the inter-prediction module 121 performs an IME on the PU, the inter-prediction module 121 can search for one or more reference pictures for the reference block for the PU. After discovering the reference block for the PU, the inter-prediction module 121 can generate a motion vector indicating the spatial displacement between the PU and the reference block for the PU with integer accuracy. When the inter-prediction module 121 executes the FME on the PU, the inter-prediction module 121 can improve the motion vector generated by executing the IME on the PU. The motion vector generated by running FME on the PU may have sub-integer accuracy (eg, 1/2 pixel accuracy or 1/4 pixel accuracy). After generating the motion vector of the PU, the inter-prediction module 121 can generate the predicted picture block of the PU by using the motion vector of the PU.

In some feasible implementations in which the inter-prediction module 121 signals the PU motion information to the decoder in AMVP mode, the inter-prediction module 121 may generate a list of candidate predicted motion vectors for the PU. can. The list of predicted motion vectors of candidates is the predicted motion vector of one or more original candidates and the predicted motion vector of one or more additional candidates derived from the predicted motion vector of one or more original candidates. It may also include a motion vector. After generating a list of candidate predicted motion vectors for the PU, the inter-prediction module 121 selects the candidate predicted motion vectors from the list of candidate predicted motion vectors and for the PU. Motion vector differences (MVDs) can be generated. The MVD for the PU can indicate the difference between the motion vector represented by the predicted motion vector of the selected candidate and the motion vector generated for the PU by the IME and FME. In these feasible implementations, the inter-prediction module 121 indexes the candidate's predicted motion vectors to identify the position of the selected candidate's predicted motion vector in the list of candidate's predicted motion vectors. Can be output. The inter-prediction module 121 can further output the MVD of the PU. Hereinafter, a feasible implementation of the advanced motion vector prediction (AMVP) mode will be described in detail in FIG. 11 in the embodiment of the present application.

In addition to executing IME and FME on the PU to generate motion information on the PU, the inter-prediction module 121 may further perform a merge operation on the PU. When the inter-prediction module 121 performs a merge operation on the PU, the inter-prediction module 121 can generate a list of candidate predicted motion vectors for the PU. The list of candidate predicted motion vectors for PU is one or more original candidate predicted motion vectors and one or more additions derived from one or more original candidate predicted motion vectors. It may include a candidate's predicted motion vector. The predicted motion vector of one or more original candidates in the list of predicted motion vectors of the candidate includes the predicted motion vector of one or more spatial candidates and the predicted motion vector of the time candidate. Can be done. The predicted motion vector of the spatial candidate can indicate motion information for another PU in the current picture. The predicted motion vector of the time candidate may be based on the motion information of the corresponding PU in a picture different from the current picture. The predicted motion vector of a time candidate is sometimes referred to as a time motion vector prediction (TMVP).

After generating the list of candidate predicted motion vectors, the inter-prediction module 121 can select one candidate predicted motion vector from the list of candidate predicted motion vectors. Next, the inter-prediction module 121 can generate a predictive picture block of the PU based on the reference block indicated by the motion information of the PU. In merge mode, the motion information of the PU may be the same as the motion information indicated by the predicted motion vector of the selected candidate. FIG. 10 is a flowchart of an example of the merge mode.

After the IME and FME generate the predicted picture block of the PU and the merge operation to generate the predicted picture block of the PU, the inter-prediction module 121 may generate the predicted picture block or the predicted picture block generated by performing the FME operation. You can select the predicted picture blocks generated by performing the merge operation. In some feasible implementations, the inter-prediction module 121 has a rate distortion cost between the predicted picture block generated by performing the FME operation and the predicted picture block generated by performing the merge operation. By analyzing the, the predicted picture block of the PU can be selected.

After the inter-prediction module 121 selects the predicted picture block of the PU generated by partitioning the current CU according to each partitioning pattern (in some implementations, the coding tree unit CTU divides it into CUs). After that, the CU is not further subdivided into smaller PUs, in which case the PU is equivalent to the CU), the inter-prediction module 121 can select the partitioning pattern for the current CU. .. In some implementations, the inter-prediction module 121 analyzes the current rate distortion cost of the selected predicted picture block of the PU generated by partitioning the current CU according to each partitioning pattern. You can select the partitioning pattern for the CU. The inter-prediction module 121 can output the prediction picture block associated with the PU belonging to the selected partitioning pattern to the residual generation module 102. The inter-prediction module 121 can output the syntax element of the motion information of the PU belonging to the selected partitioning pattern to the entropy coding module 116.

In the schematic shown in FIG. 4, the inter-prediction module 121 includes IME modules 180A to 180N (collectively referred to as “IME module 180”), FME modules 182A to 182N (collectively referred to as “FME module 182”), and modules 184A to 184N. (Collectively referred to as "merge module 184"), PU pattern decision modules 186A to 186N (collectively referred to as "PU pattern decision module 186"), and CU pattern decision module 188 (CTU vs. CU pattern decision process). May be further executed).

The IME module 180, FME module 182, and merge module 184 can perform IME operations, FME operations, and merge operations on the PU of the current CU, respectively. In the schematic shown in FIG. 4, the inter-prediction module 121 is described as including a separate IME module 180, a separate FME module 182, and a separate merge module 184 for each PU in each partitioning pattern of the CU. Will be done. In another feasible implementation, the inter-prediction module 121 does not include a separate IME module 180, a separate FME module 182, or a separate merge module 184 for each PU in each partitioning pattern of the CU.

As shown in the schematic of FIG. 4, the IME module 180A, the FME module 182A, and the merge module 184A each have an IME for a PU generated by partitioning the CU according to a 2N × 2N partitioning pattern. Operations, FME operations, and merge operations can be performed. The PU mode decision module 186A can select one of the predictive picture blocks generated by the IME module 180A, the FME module 182A, and the merge module 184A.

The IME module 180B, FME module 182B, and merge module 184B each perform IME, FME, and merge operations on the left PU generated by partitioning the CU according to an N × 2N partitioning pattern. Can be executed. The PU mode decision module 186B can select one of the predictive picture blocks generated by the IME module 180B, the FME module 182B, and the merge module 184B.

The IME module 180C, FME module 182C, and merge module 184C each perform IME, FME, and merge operations on the right PU generated by partitioning the CU according to the N × 2N partitioning pattern. Can be executed. The PU mode decision module 186C can select one of the predictive picture blocks generated by the IME module 180C, the FME module 182C, and the merge module 184C.

The IME module 180N, the FME module 182N, and the merge module 184, respectively, perform an IME operation, an FME operation, and a merge operation on the lower right PU generated by partitioning the CU according to the N × N partitioning pattern. Can be executed. The PU mode decision module 186N can select one of the predictive picture blocks generated by the IME module 180N, the FME module 182N, and the merge module 184N.

The PU pattern decision module 186 selects a predictive picture block by analyzing the rate distortion cost of multiple possible predictive picture blocks and provides the optimal rate distortion cost for a given decoding scenario. -You can select a block. For example, for applications with limited bandwidth, the PU mode decision module 186 may prefer predictive picture blocks with increased compression ratios, and for other applications PU mode. The decision module 186 may prefer predictive picture blocks that improve the quality of the reconstructed video. After the PU pattern decision module 186 selects the predicted picture block for the PU of the current CU, the CU pattern decision module 188 selects the partitioning pattern for the current CU to the selected partitioning pattern. Outputs the predicted picture block and motion information for the PU to which it belongs.

FIG. 5 is a schematic diagram of an example of a coding unit and adjacent picture blocks associated with the coding unit according to an embodiment of the present application, with predicted movements of the CU 250 and exemplary candidates associated with the CU 250. It is a schematic diagram which shows the vector positions 252A to 252E. In the present application, the candidate's predicted motion vector positions 252A to 252E may be collectively referred to as the candidate's predicted motion vector position 252. The candidate's predicted motion vector position 252 represents the predicted motion vector of the spatial candidate in the same picture as the CU 250. The candidate's predicted motion vector position 252A is to the left of the CU250. The candidate's predicted motion vector position 252B is above the CU250. The candidate's predicted motion vector position 252C is in the upper right corner of the CU250. The candidate's predicted motion vector position 252D is at the bottom left of the CU250. The candidate's predicted motion vector position 252E is at the top left of the CU250. FIG. 8 shows an implementation example of how the inter-prediction module 121 and motion compensation module 162 can generate a list of candidate predicted motion vectors. Hereinafter, the implementation will be described in relation to the inter-prediction module 121. However, it should be understood that the motion compensation module 162 may implement the same technique and thus generate a list of predicted motion vectors of the same candidate. In this embodiment of the present application, the picture block in which the predicted motion vector position of the candidate is located is referred to as a reference block. Further, the reference block includes a spatial reference block, eg, a picture block in which 252A to 252E are located, and a time reference block, eg, a picture block in which a co-located block is located, or is co-located. Contains picture blocks that are spatially adjacent to each other.

FIG. 6 is an exemplary flow chart for constructing a list of candidate predicted motion vectors according to an embodiment of the present application. Although the technique of FIG. 6 is described on the basis of a list containing five candidate predicted motion vectors, the technique described herein may optionally be used with a list of different sizes. Each of the five candidate predicted motion vectors can have an index (eg, 0-4). The technique of FIG. 6 will be described on the basis of a typical video decoder. A typical video decoder may be, for example, a video encoder (eg, video encoder 20) or a video decoder (eg, video decoder 30).

In order to reconstruct the list of predicted motion vectors of the candidates according to the implementation of FIG. 6, the video decoder first considers four spatial candidates (602), and each spatial candidate has one predicted. Corresponding to the motion vector, the predicted motion vector of the four spatial candidates may include the candidate's predicted motion vector positions 252A, 252B, 252C, 252D. The predicted motion vectors of the four spatial candidates correspond to the motion information of the four PUs in the same picture as the current CU (eg, CU250). In other words, the predicted motion vectors of the four spatial candidates are the predicted motion vectors of the four PUs. In this embodiment of the present application, the video decoder must first determine the four PUs when acquiring the predicted motion vectors of the four spatial candidates. For example, the process by which the video decoder determines one of the four PUs may be as follows: if the PU and the current CU are in the same CTB or CTB row range: , The unupdated initial motion vector of the PU is used as the predicted motion vector of the spatial candidate corresponding to the PU, or the PU and the current CU are not in the same CTB or CTB row range. The updated final motion vector of the PU is used as the predicted motion vector of the spatial candidates corresponding to the PU. The updated final motion vector for the PU can be obtained from the target storage space where the PU is stored. In this case, the target storage space stores the updated final motion vector. In some embodiments, the target storage space stores the MVD. In this case, the updated final motion vector for the PU may instead be obtained based on the MVD stored in the target storage space and the unupdated initial motion vector for the PU. good.

The video decoder can consider the predicted motion vectors of the four spatial candidates in the list in a specified order. For example, the candidate's predicted motion vector position 252A may be considered first. If the candidate predicted motion vector position 252A is available, the candidate predicted motion vector position 252A may be assigned to index 0. If the candidate predicted motion vector position 252A is not available, the video decoder may not add the candidate predicted motion vector position 252A to the list of candidate predicted motion vectors. The candidate's predicted motion vector position may not be available for a variety of reasons. For example, if the candidate's predicted motion vector position is not in the current picture, the candidate's predicted motion vector position may not be available. In another feasible implementation, if the candidate's predicted motion vector position goes through intra-prediction, the candidate's predicted motion vector position may not be available. In another feasible implementation, if the candidate's predicted motion vector position is in a different slice than the slice corresponding to the current CU, the candidate's predicted motion vector position may not be available. be.

After considering the candidate predicted motion vector position 252A, the video decoder may consider the candidate predicted motion vector position 252B. If the candidate's predicted motion vector position 252B is available and different from the candidate's predicted motion vector position 252A, the video decoder will place the candidate's predicted motion vector position 252B as the candidate's predicted motion vector position 252B. It may be added to the list of motion vectors. In this particular context, the terms "same" or "different" mean that a candidate's predicted motion vector position is associated with the same or different motion information. Therefore, if the predicted motion vector positions of the two candidates have the same motion information, then the predicted motion vector positions of the two candidates are considered to be the same, or the predicted motion vector positions of the two candidates are If they have different motion information, the predicted motion vector positions of the two candidates are considered to be different. If the candidate predicted motion vector position 252A is not available, the video decoder may assign the candidate predicted motion vector position 252B to index 0. If a candidate predicted motion vector position 252A is available, the video decoder may assign the candidate predicted motion vector position 252 to index 1. If the candidate's predicted motion vector position 252B is unavailable or is the same as the candidate's predicted motion vector position 252A, the video decoder skips the candidate's predicted motion vector position 252B. , Do not add the candidate's predicted motion vector position 252B to the list of candidate's predicted motion vectors.

Similarly, the video decoder takes into account the candidate's predicted motion vector position 252C and decides whether to add the candidate's predicted motion vector position 252C to the list. If a candidate predicted motion vector position 252C is available and is different from the candidate predicted motion vector positions 252B and 252A, the video decoder will use the candidate predicted motion vector position 252C as the next available. Can be assigned to various indexes. If the candidate's predicted motion vector position 252C is not available or is identical to at least one of the candidate's predicted motion vector positions 252A and 252B, the video decoder will perform the candidate's predicted motion vector. Do not add position 252C to the list of candidate predicted motion vectors. The video decoder then considers the candidate predicted motion vector position 252D. If a candidate predicted motion vector position 252D is available and is different from the candidate predicted motion vector positions 252A, 252B, and 252C, the video decoder sets the candidate predicted motion vector position 252D to the next. Can be assigned to the available indexes of. If the candidate's predicted motion vector position 252D is not available or is identical to at least one of the candidate's predicted motion vector positions 252A, 252B, and 252C, the video decoder predicts the candidate. The motion vector position 252D is not added to the list of candidate predicted motion vectors. In the above implementation, the candidate predicted motion vectors 252A to 252D are considered to determine whether the candidate predicted motion vectors 252A to 252D are added to the list of candidate predicted motion vectors. An example of how to do this is outlined. However, in some implementations, all of the candidate predicted motion vectors 252A to 252D may first be added to the list of candidate predicted motion vectors, and then iterative candidate predicted motion. The vector position is removed from the list of candidate predicted motion vectors.

After the video decoder considers the predicted motion vectors of the first four spatial candidates, the list of predicted motion vectors of the candidates may contain the predicted motion vectors of the four spatial candidates, or It may contain predicted motion vectors of less than four spatial candidates. If the list contains the predicted motion vectors of four spatial candidates (yes at 604), i.e., the video decoder considers the time candidates (606), and each time candidate is one. Corresponds to the predicted motion vector of one candidate. The predicted motion vector of the time candidate may correspond to the motion information of the PU at the same position in the picture different from the current picture. If the predicted motion vector of the time candidate is available and different from the predicted motion vector of the first four spatial candidates, the video decoder assigns the predicted motion vector of the time candidate to index 4. If the predicted motion vector of the time candidate is not available or is identical to one of the predicted motion vectors of the first four spatial candidates, the video decoder will use the predicted motion vector of the time candidate. Is not added to the list of candidate predicted motion vectors. Thus, after the video decoder considers the predicted motion vectors of the time candidates (606), the list of predicted motion vectors of the candidates is the first of the five predicted motion vectors considered in 602. It may contain the predicted motion vectors of the four spatial candidates and the predicted motion vectors of the time candidates considered in 604, or the first predicted motion vectors of the four candidates considered in 602. Predicted motion vectors of the four spatial candidates). If the list of candidate predicted motion vectors contains five candidate predicted motion vectors (yes in 608), i.e., the video decoder completes the construction of the list.

If the list of candidate predicted motion vectors contains four candidate predicted motion vectors (No in 608), the video decoder may consider the fifth spatial candidate's predicted motion vector. You can (610). The predicted motion vector of the fifth spatial candidate may (eg) correspond to the predicted motion vector position 252E of the candidate. If a candidate predicted motion vector corresponding to position 252E is available and is different from the candidate predicted motion vector corresponding to positions 252A, 252B, 252C, and 252D, the video decoder is a fifth. The predicted motion vector of the spatial candidate can be added to the list of predicted motion vectors of the candidate, and the predicted motion vector of the fifth spatial candidate can be assigned to the index 4. If the candidate's predicted motion vector corresponding to position 252E is not available, or if the candidate's predicted motion vector is identical to the candidate's predicted motion vector corresponding to positions 252A, 252B, 252C, and 252D. In some cases, the video decoder may not add the candidate predicted motion vector corresponding to position 252 to the list of candidate predicted motion vectors. Therefore, after the predicted motion vectors of the fifth spatial candidate are considered (610), the list predicts the predicted motion vectors of the five candidates (the first four spatial candidates considered in 602). It may contain motion vectors, and the predicted motion vectors of the fifth spatial candidate considered in 610, or the predicted motion vectors of four candidates (the first four spatial candidates examined in 602). Predicted motion vector) may be included.

If the list of candidate predicted motion vectors contains 5 candidates (yes in 612), that is, if it contains 5 candidate predicted motion vectors, the video decoder will list the candidate predicted motion vectors. Completes the generation of. If the list of candidate predicted motion vectors contains four candidate predicted motion vectors (No at 612), the video decoder will run until the list contains five candidates (yes at 616), i.e. five. Add the manually generated candidate predicted motion vectors until they contain the candidate predicted motion vectors (614).

If the list contains less than four predicted motion vectors of spatial candidates (No in 604) after the video decoder considers the predicted motion vectors of the first four spatial candidates, then the video decoder is the first. The 5 spatial candidates can be considered (618), i.e. the predicted motion vector of the 5th spatial candidate can be considered. The predicted motion vector of the fifth spatial candidate may (eg) correspond to the predicted motion vector position 252E of the candidate. If the candidate predicted motion vector corresponding to position 252E is available and is different from the candidate predicted motion vector present in the list of candidate predicted motion vectors, the video decoder is in the fifth space. Candidate predicted motion vectors can be added to the list of candidate predicted motion vectors and the fifth spatial candidate predicted motion vector can be assigned to the next available index. If the candidate predicted motion vector corresponding to position 252E is not available or is identical to one of the candidate predicted motion vectors present in the candidate's list of predicted motion vectors, video. The decoder may not add the candidate predicted motion vector corresponding to position 252E to the list of candidate predicted motion vectors. The video decoder can then consider temporal candidates, i.e., temporally predicted motion vectors (620). If the predicted motion vector of the time candidate is available and different from the predicted motion vector of the candidate present in the list of predicted motion vectors of the candidate, the video decoder will display the predicted motion vector of the time candidate. Can be added to the list of candidate predicted motion vectors and the time candidate predicted motion vector can be assigned to the next available index. If the predicted motion vector of the time candidate is unavailable or is identical to one of the predicted motion vectors of the candidate present in the list of predicted motion vectors of the candidate, the video decoder will run the time. Candidate predicted motion vectors may not be added to the list of candidate predicted motion vectors.

After the predicted motion vector (618) of the fifth space candidate and the predicted motion vector (620) of the time candidate are considered, the list of predicted motion vectors of the candidate is the fifth space candidate. If it contains a predicted motion vector (yes at 622), the video decoder completes the generation of a list of candidate predicted motion vectors. If the list of candidate predicted motion vectors contains less than 5 candidate predicted motion vectors (No at 622), the video decoder will continue until the list contains 5 candidate predicted motion vectors. (Yes at 616), add the predicted motion vector of the manually generated candidate (614).

According to the technique of the present application, the predicted motion vectors of additional merge candidates can be manually generated after the predicted motion vectors of the spatial candidates and the predicted motion vectors of the time candidates. As a result, the size of the list of predicted motion vectors of the merge candidates will always be the predicted motion vectors of the specified amount of merge candidates (eg, the predictions of the five candidates in the feasible implementation of FIG. 6 above. Is equal to the motion vector). The predicted motion vectors of the additional merge candidates are, for example, the predicted motion vectors of the combined bidirectional predictive merge candidates (candidate predicted motion vector 1), the scaled bidirectional predictive merge candidate predictions. It may include a motion vector (candidate predicted motion vector 2) and a zero vector merge candidate predicted motion vector (candidate predicted motion vector 3). See FIGS. 7-9 for a specific description of the above three cases of predicted motion vectors for additional merge candidates.

FIG. 7 is a schematic diagram of an example of adding a combined candidate motion vector to a list of predicted motion vectors of merge mode candidates according to an embodiment of the present application. The predicted motion vectors of the combined bi-predictive merge candidates can be generated by combining the predicted motion vectors of the original merge candidates. Specifically, the predicted motion vectors of the two original candidates (having mvL0 and refIdxL0, or mvL1 and refIdxL1) can be used to generate the predicted motion vectors of the bipredictive merge candidates. In FIG. 7, the predicted motion vectors of the two candidates are included in the list of predicted motion vectors of the original merge candidates. The prediction type of the predicted motion vector of one candidate is one-way prediction by using List 0, and the prediction type of the predicted motion vector of the other candidate is one-way prediction by using List 1. be. In this feasible implementation, mvL0_A and ref0 are obtained from Listing 0, and mvL1_B and ref0 are obtained from Listing 1. It is then possible to generate the predicted motion vectors of the bi-predictive merge candidates (having mvL0_A and ref0 in Listing 0 and mvL1_B and ref0 in Listing 1), and the predicted motion vectors of the bi-predictive merge candidates. Is checked for different from the candidate's predicted motion vector present in the candidate's list of predicted motion vectors. If the predicted motion vector of a bi-predicted merge candidate is different from the predicted motion vector of an existing candidate, the video decoder will display the predicted motion vector of the bi-predicted merge candidate as a list of the predicted motion vectors of the candidate. Can be added to.

FIG. 8 is a schematic diagram of an example of adding a scaled candidate motion vector to a list of predicted motion vectors of merge mode candidates according to an embodiment of the present application. The predicted motion vector of the scaled bi-predictive merge candidate may be generated by scaling the predicted motion vector of the original merge candidate. Specifically, one original candidate predicted motion vector (with mvLX and refIdxLX) can be used to generate a bipredicted merge candidate predicted motion vector. In the feasible implementation of FIG. 8, the predicted motion vectors of the two candidates are included in the list of predicted motion vectors of the original merge candidates. The prediction type of the predicted motion vector of one candidate is one-way prediction by using List 0, and the prediction type of the predicted motion vector of the other candidate is one-way prediction by using List 1. be. In this feasible implementation, mvL0_A and ref0 are obtained from Listing 0, ref0 is copied to Listing 1 and shown as the reference index ref0'. MvL0'_A can then be calculated by scaling mvL0_A with ref0 and ref0'. Scaling may depend on the POC (Picture Order Count, POC) distance. It is then possible to generate predicted motion vectors of bi-predictive merge candidates (having mvL0_A and ref0 in Listing 0, mvL0'_A and ref0' in List 1) and predicted bi-predictive merge candidates. It is checked whether the motion vector is repeated. If the predicted motion vector of the bi-predicted merge candidate is not repeated, it is possible that the predicted motion vector of the bi-predicted merge candidate is added to the list of predicted motion vectors of the merge candidate.

FIG. 9 is a schematic diagram of an example of adding a zero motion vector to a list of predicted motion vectors of merge mode candidates according to an embodiment of the present application. The predicted motion vector of the zero vector merge candidate can be generated by combining the zero vector and the reference index that can be referenced. If the predicted motion vector of the zero vector candidate is not repeated, the predicted motion vector of the zero vector candidate can be added to the list of predicted motion vectors of the merge candidate. The motion information of the predicted motion vector of each generated merge candidate can be compared with the motion information of the predicted motion vector of the previous candidate in the list.

In a feasible implementation, if the predicted motion vector of the newly generated candidate is different from the predicted motion vector of the candidate present in the list of predicted motion vectors of the candidate, the predicted motion vector of the generated candidate is predicted. The motion vector is added to the list of predicted motion vectors for merge candidates. The process of determining whether a candidate's predicted motion vector differs from the candidate's predicted motion vector present in the candidate's list of predicted motion vectors is often referred to as pruning. Pruning allows each newly generated predicted motion vector of a candidate to be compared with the predicted motion vector of an existing candidate in the list. In some feasible implementations, the pruning process compares the predicted motion vector of one or more new candidates with the predicted motion vector of the candidate present in the list of predicted motion vectors of the candidate. It is possible to include skipping the addition of new candidate predicted motion vectors that are identical to the candidate's predicted motion vectors present in the candidate's list of predicted motion vectors. In some other feasible implementation, the pruning process adds one or more new candidate predicted motion vectors to the list of candidate predicted motion vectors and then iterative candidate predictions. May include excluding motion vectors from the list.

In a feasible implementation of the present application, a method of predicting motion information of a picture block to be processed during interprediction is: at least one picture in which the motion vector is determined in the picture in which the picture block to be processed is located. At least one picture block in which the motion vector is determined in the step of acquiring the motion information of the block is a picture block not adjacent to the picture block to be processed, and the motion vector is determined. A step including a picture block and a step of acquiring the first identification information, wherein the first identification information determines the target motion information in the motion information of at least one picture block in which the motion vector is determined. Includes a step used in the process and a step of predicting the motion information of the picture block to be processed based on the target motion information.

FIG. 10 is a flowchart of an example of the merge mode according to the embodiment of the present application. The video encoder (eg, the video encoder 20) can perform the merge operation 200. In another feasible implementation, the video encoder may perform a different merge operation than the merge operation 200. For example, in another feasible implementation, the video encoder can perform the merge operation, and the video encoder is different from more or less steps in the merge operation 200, or steps in the merge operation 200. Perform the steps. In another feasible implementation, the video encoder can perform the steps of the merge operation 200 in different orders or in parallel. The encoder may further perform merge operation 200 on the PU encoded in skip mode.

After the video encoder initiates the merge operation 200, the video encoder can generate a list of candidate predicted motion vectors for the current PU (202). The video encoder can generate a list of candidate predicted motion vectors for the current PU in a variety of ways. For example, a video encoder can generate a list of candidate predicted motion vectors for the current PU according to one of the technical examples described below in connection with FIGS. 7-9.

As mentioned above, the list of predicted motion vectors for the current PU candidates may include the predicted motion vectors for the time candidates. The predicted motion vector of the time candidate can indicate the motion information of the PUs at the same position in time. PUs at the same position may be spatially located in the picture frame at the same position as the current PU, but may be located in the reference picture instead of the current picture. In the present application, a reference picture including a corresponding temporal PU may be referred to as a related reference picture. In this application, the reference picture index of the associated reference picture may be referred to as the associated reference picture index. As mentioned above, the current picture may be associated with one or more reference picture lists (eg, List 0 and List 1). The reference picture index can indicate a reference picture by indicating the position of the reference picture in the reference picture list. In some feasible implementations, the current picture may be associated with a combined reference picture list.

For some video encoders, the associated reference picture index is the PU's reference picture index that covers the reference index source position associated with the current PU. In these video encoders, the reference index source located relative to the current PU is adjacent to the left of the current PU or above the current PU. In the present application, if the picture block associated with the PU contains a particular location, the PU may "cover" the particular location. For these video encoders, the video encoder may use the reference picture index 0 if the source position of the reference index is not available.

In some examples, the reference index source location associated with the current PU is within the current CU. In these examples, a PU that covers the reference index source position associated with the current PU may be considered available if the PU is above or to the left of the current CU. However, the video encoder may need to access motion information for another PU in the current CU to determine a reference picture that includes a PU in the same position. Thus, these video encoders can use motion information (eg, reference picture index) for PUs belonging to the current CU to generate predicted motion vectors for time candidates for the current PU. In other words, these video encoders can use the motion information of PUs belonging to the current CU to generate predicted motion vectors of temporal candidates. Therefore, the video encoder may not be able to generate a list of candidate predicted motion vectors for the current PU and the PU covering the reference index source position associated with the current PU in parallel. No.

According to the technique of the present application, the video encoder can explicitly set the associated reference picture index without reference to the reference picture index of any other PU. In this way, the video encoder can generate a list of candidate predicted motion vectors in parallel for the current PU and another PU of the current CU. Since the video encoder explicitly sets the associated reference picture index, the associated reference picture index is not based on the motion information of any other PU in the current CU. In some feasible implementations where the video encoder explicitly sets the associated reference picture index, the video encoder sets the associated reference picture index to a fixed, pre-set reference picture. -It can always be set to the index (for example, 0). In this way, the video encoder generates a predicted motion vector of temporal candidates based on the motion information of the co-located PUs in the reference frame indicated by the preset reference picture index. It is possible to add the predicted motion vectors of the temporal candidates to the list of predicted motion vectors of the current CU candidates. The list of candidate predicted motion vectors is also a reference frame list.

In a feasible implementation where the video encoder explicitly sets the associated reference picture index, the video encoder has a syntax structure (eg, picture header, slice header, APS, or other syntax). In the structure), the associated reference picture index can be explicitly signaled. In this feasible implementation, the video encoder signals the decoder with a relevant reference picture index for each LCU (ie, CTU), CU, PU, TU, or another type of subblock. Can be done. For example, the video encoder can signal that the associated reference picture index for each PU of the CU is equal to "1".

In some feasible implementations, the associated reference picture index may be set implicitly rather than explicitly. In these feasible implementations, the video encoder is indicated by a PU reference picture index that covers the position outside the current CU, even if these positions are not exactly adjacent to the current PU. By using the motion information for the PU in the reference picture, it is possible to generate a predicted motion vector for each temporal candidate in the list of predicted motion vectors for the candidate for the PU of the current CU.

After generating a list of candidate predicted motion vectors for the current PU, the video encoder generates a predictive picture block associated with the candidate's predicted motion vector in the list of candidate predicted motion vectors. Can be (204). The video encoder determines the current PU motion information based on the motion information of the indicated candidate's predicted motion vector, and then generates a predictive picture block associated with the candidate's predicted motion vector. Therefore, it is possible to generate a predictive picture block based on one or more reference blocks indicated by the motion information of the current PU. The video encoder can then select one candidate predicted motion vector from the list of candidate predicted motion vectors (206). The video encoder can select candidate predicted motion vectors in various ways. For example, the video encoder can select one candidate predicted motion vector by analyzing the rate distortion cost of each predicted picture block associated with the candidate predicted motion vector.

After selecting a candidate's predicted motion vector, the video encoder can output the candidate's predicted motion vector index (208). The candidate's predicted motion vector index is also the reference frame list index, and the candidate's predicted motion vector index is the candidate's predicted motion vector index selected from the candidate's list of predicted motion vectors. The position of the motion vector can be shown. In some feasible implementations, the candidate's predicted motion vector index may be expressed as "merge_idx".

FIG. 11 is a flowchart of an example of the advanced motion vector prediction mode according to the embodiment of the present application. The video encoder (eg, video encoder 20) can perform AMVP operation 210.

After the video encoder initiates the AMVP operation 210, the video encoder can generate one or more motion vectors for the current PU (211). The video encoder can perform integer motion estimation and fractional motion estimation to generate motion vectors for the current PU. As mentioned above, the current picture may be associated with two reference picture lists (List 0 and List 1). If one-way prediction is performed on the current PU, the video encoder can generate a List-0 motion vector or a List-1 motion vector for the current PU. The list-0 motion vector can indicate the spatial displacement between the picture block for the current PU and the reference block of the reference picture in list 0. The Listing-1 motion vector can indicate the spatial displacement between the picture block for the current PU and the reference block of the reference picture in Listing 1. If bidirectional prediction is performed on the current PU, the video encoder can generate a List-0 motion vector and a List-1 motion vector for the current PU.

After generating one or more motion vectors for the current PU, the video encoder may generate a predictive picture block for the current PU (212). The video encoder can generate a predictive picture block for the current PU based on one or more reference blocks indicated by one or more motion vectors for the current PU.

In addition, the video encoder can generate a list of candidate predicted motion vectors for the current PU (213). The video decoder can generate a list of candidate predicted motion vectors for the current PU in a variety of ways. For example, a video encoder may generate a list of candidate predicted motion vectors for the current PU according to one or more feasible implementations described below in connection with FIGS. 6-9. Can be done. In some feasible implementations, if the video encoder produces a list of candidate predicted motion vectors in AMVP operation 210, the list of candidate predicted motion vectors will be two candidate predicted motion vectors. May be limited to motion vectors. In contrast, if the video encoder produces a list of candidate predicted motion vectors in a merge operation, the list of candidate predicted motion vectors will be a list of more candidate predicted motion vectors (eg,). It may contain 5 candidate predicted motion vectors).

After generating a list of candidate predicted motion vectors for the current PU, the video encoder will in the list of candidate predicted motion vectors one or more for each candidate's predicted motion vector. A motion vector difference (MVD) can be generated (214). The video encoder determines the difference between the motion vector indicated by the candidate's predicted motion vector and the corresponding motion vector for the current PU to generate the motion vector difference for the candidate's predicted motion vector. can do.

If one-way prediction is performed on the current PU, the video encoder can generate a single MVD for each candidate predicted motion vector. If bidirectional prediction is performed on the current PU, the video encoder can generate two MVDs for each candidate predicted motion vector. The first MVD can show the difference between the motion vector represented by the candidate predicted motion vector and the list-0 motion vector for the current PU. The second MVD can show the difference between the motion vector represented by the candidate's predicted motion vector and the List-1 motion vector for the current PU.

The video encoder can select one or more candidate predicted motion vectors from the list of candidate predicted motion vectors (215). The video encoder can select one or more candidate predicted motion vectors in various ways. For example, the video encoder can select a candidate predicted motion vector that matches the associated motion vector of the motion vector to be encoded with minimal error. This can reduce the amount of bits required to represent the motion vector difference for the candidate's predicted motion vector.

After selecting one or more candidate predicted motion vectors, the video encoder will have one or more reference picture indexes for the current PU and one or more candidate predicted motion vectors for the current PU. It is possible to output the index and one or more motion vector differences with respect to the predicted motion vector of one or more selected candidates (216).

In an example where the current picture is associated with two reference picture lists (List 0 and List 1) and one-way prediction is performed for the current PU, the video encoder has a reference picture index to List 0 (List 0). A reference picture index (“ref_idx_11”) for “ref_idx_10”) or Listing 1 can be output. The video encoder also provides a candidate predicted motion vector that indicates the position of the selected candidate's predicted motion vector for the list-0 motion vector with respect to the current PU within the list of candidate predicted motion vectors. -The index ("mbp_10_flag") can be output. Alternatively, the video encoder is a candidate's predicted motion vector that indicates the position of the selected candidate's predicted motion vector for the list-1 motion vector with respect to the current PU within the list of candidate's predicted motion vectors. A motion vector index ("mvp_11_flag") can be output. The video encoder can also output a List-0 motion vector for the current PU or an MVD for the List-1 motion vector.

In an example where the current picture is associated with two reference picture lists (List 0 and List 1) and bidirectional prediction is performed for the current PU, the video encoder is for List 0. A reference picture index (“ref_idx_11”) can be output for the reference picture index (“ref_idx_10”) and list 1. In addition, the video encoder indicates the position of the selected candidate's predicted motion vector for the list-0 motion vector with respect to the current PU in the list of candidate's predicted motion vectors. The index ("mvp_10_flag") can be output. In addition, the video encoder indicates the position of the selected candidate's predicted motion vector of the list-1 motion vector with respect to the current PU in the list of candidate's predicted motion vectors. The index ("mvp_11_flag") can be output. The video encoder can also output an MVD for the List-0 motion vector for the current PU and an MVD for the List-1 motion vector for the current PU.

FIG. 12 is a flowchart of an example of motion compensation performed by a video decoder (eg, video decoder 30) according to an embodiment of the present application.

If the video decoder performs a motion compensation operation 220, the video decoder can be instructed by the predicted motion vector of the selected candidate for the current PU (222). For example, the video decoder receives a candidate predicted motion vector index indicating the position of the selected candidate's predicted motion vector in the list of candidate predicted motion vectors for the current PU. be able to.

If the motion information of the current PU is encoded in AMVP mode and bidirectional prediction is performed for the current PU, the video decoder will have the predicted motion vector index of the first candidate and the second candidate. It is possible to receive the predicted motion vector index. The predicted motion vector index of the first candidate indicates the position of the predicted motion vector of the selected candidate of the list-0 motion vector with respect to the current PU in the list of predicted motion vectors of the candidate. The predicted motion vector index of the second candidate indicates the position of the predicted motion vector of the selected candidate of the list-1 motion vector with respect to the current PU in the list of predicted motion vectors of the candidate. In some feasible implementations, a single syntax element can be used to identify the predicted motion vector indexes of the two candidates.

In addition, the video decoder can generate a list of candidate predicted motion vectors for the current PU (224). The video decoder can generate a list of candidate predicted motion vectors for the current PU in a variety of ways. For example, a video decoder can generate a list of candidate predicted motion vectors for the current PU by using the techniques described below in connection with FIGS. 5-9. If the video decoder produces a candidate predicted motion vector in time for a list of candidate predicted motion vectors, the video decoder will be co-located PU as described above with respect to FIG. A reference picture index that identifies the reference picture containing the can be set explicitly or implicitly.

After generating a list of candidate predicted motion vectors for the current PU, the video decoder is indicated by one or more candidate predicted motion vectors in the list of candidate predicted motion vectors for the current PU. Based on the motion information, the motion information for the current PU can be determined (225). For example, if the current PU motion information is encoded in merge mode, the current PU motion information may be the same as the motion information indicated by the predicted motion vector of the selected candidate. If the current PU motion information is encoded in AMVP mode, the video decoder will have one or more motion vectors represented by the predicted motion vectors of the selected candidates and one or more represented by the bitstream. By using with MVD, one or more motion vectors for the current PU can be reconstructed. The current PU reference picture index and predictive direction identifier are identical to one or more reference picture indexes and one or more predictive direction identifiers of one or more selected candidate predicted motion vectors. You may. After determining the motion information for the current PU, the video decoder determines the predicted picture block for the current PU based on one or more reference blocks indicated by the motion information for the current PU. Can be generated (226).

FIG. 13 is an example flowchart for updating the motion vector during video coding according to the embodiment of the present application. The picture block to be processed is a block to be encoded.

1301: Acquires the initial motion vector of the picture block to be processed based on the predicted motion vector of the picture block to be processed.

In a viable implementation, for example, in merge mode, the predicted motion vector of the picture block to be processed is used as the initial motion vector of the picture block to be processed.

The predicted motion vector of the picture block to be processed is the method shown in FIGS. 6 to 9 in the embodiment of the present application, or H. It can be acquired by any existing predicted motion vector acquisition method in the 265 standard or JEM reference mode. This is not limited. The motion vector difference can be acquired in relation to the picture block to be processed. Motion estimation is performed within a search range determined based on the predicted motion vector of the picture block to be processed, and the motion vector of the picture block to be processed and the picture to be processed obtained after motion estimation. The difference between the block and the predicted motion vector is used as the motion vector difference.

During bidirectional prediction, this step specifically: Get the first initial motion vector of the picture block to be processed based on the forward predicted motion vector of the picture block to be processed. And the step of acquiring a second initial motion vector of the picture block to be processed based on the backward predicted motion vector of the picture block to be processed.

1302: Obtains a predicted block of picture blocks to be processed based on the initial motion vector and one or more preset motion vector offsets. Details are as follows:

S13021: The picture block represented by the initial motion vector of the picture block to be processed is acquired from the reference frame of the picture block to be processed indicated by the reference frame index of the picture block to be processed, and the processing target is obtained. It is supplied as a time prediction block of the picture block of.

The initial motion vector includes a first initial motion vector and a second initial motion vector. The first initial motion vector indicates a first motion compensation block based on the first reference frame type of the picture block to be processed, and the first motion compensation block is motion compensation in the picture block to be processed. Is a reference block for executing. The second initial motion vector indicates a second motion compensation block based on the second reference frame list of the picture block to be processed, and the second motion compensation block is motion compensation in the picture block to be processed. Is a reference block for executing.

13022: To get one or more final motion vectors, add the initial motion vector of the picture block to be processed and one or more preset motion vector offsets. Here, each final motion vector indicates a search position.

The preset motion vector offset may be a preset offset vector value, a preset offset vector accuracy, or a preset offset vector range. May be. The specific value of the preset motion vector offset and the amount of the preset motion vector offset are not limited in this embodiment of the invention.

The final motion vector includes a first final motion vector and a second final motion vector. The first final motion vector indicates a first motion compensation block based on the first reference frame list of the picture block to be processed, and the first motion compensation block is motion compensation in the picture block to be processed. Is a predictive block for executing. The second final motion vector indicates a second motion compensation block based on the second reference frame list of the picture block to be processed, and the second motion compensation block is motion compensation in the picture block to be processed. Is a predictive block for executing.

13023: Acquire one or more candidate prediction blocks at one or more search positions indicated by one or more final motion vectors. Here, each search position corresponds to one candidate prediction block.

13024: From one or more candidate prediction blocks, the candidate prediction block with the smallest pixel difference from the temporal prediction block is selected as the prediction block for the picture block to be processed.

The minimum pixel difference may also be understood as the minimum distortion cost. In this case, the final motion vector of the prediction block is the candidate final motion vector corresponding to the minimum strain cost in the final motion vectors of the plurality of candidates. It should be understood that strain costs may be calculated in multiple ways. For example, the sum of the absolute differences between the pixel matrices of the candidate prediction block and the temporal prediction block may be calculated, or the sum of the absolute differences after the pixel matrix averages have been removed may be calculated. Alternatively, another relatively accurate value of the pixel matrix may be calculated. The specific content of the minimum strain cost is not limited to this embodiment of the present invention.

During bidirectional prediction, this step is specifically: the picture to be processed from the first reference frame list of the picture block to be processed indicated by the first reference frame index of the picture block to be processed. The step of acquiring the first picture block indicated by the first initial motion vector of the block, where the first picture block is the forward reference block and correspondingly the first initial motion vector is forward. From the step and the second reference frame list of the picture block to be processed indicated by the second reference frame list of the picture block to be processed, which may be the initial motion vector of. The step of retrieving the second picture block indicated by the block's second initial motion vector, where the second picture block is the back reference block and correspondingly the second initial motion vector is backward. A step that may be an initial motion vector and a step that performs weighting processing on the first picture block and the second picture block in order to acquire the time prediction block block of the picture block to be processed. To get one or more first final motion vectors, add the first initial motion vector of the picture block to be processed and one or more preset motion vector offsets. To get one or more second final motion vectors, add the second initial motion vector of the picture block to be processed and one or more preset motion vector offsets. Obtain one or more first candidate prediction blocks at one or more search positions indicated by a step and one or more first final motion vectors, and one or more second final motion vectors. A step of acquiring one or more first candidate prediction blocks at one or more search positions indicated by, and a candidate prediction block having the smallest pixel difference from the time prediction block from one or more first candidate prediction blocks. , The candidate prediction block having the smallest pixel difference from the time prediction block from one or more backward candidate prediction blocks is selected as the first prediction block of the picture block to be processed. 2 The first prediction block to acquire the step to select as the prediction block and the prediction block of the picture block to be processed. A step of executing a weighting process for the first prediction block and the first prediction block is included.

In some embodiments, another motion vector in the direction indicated by the candidate's final motion vector corresponding to the minimum strain cost in the final motion vector of the plurality of candidates is, instead, the final motion of the predicted block. It may be used as a vector. Specifically, the final motion vector of the prediction block may be selected based on the first preset condition.

In a possible implementation, the final motion vector of the plurality of candidates may be a plurality of motion vectors in multiple directions. First, a particular accuracy in each direction may be selected. For example, the final motion vector of the candidate with integer pixel accuracy is used as the first vector in each direction, and the first vector corresponding to the minimum strain cost is selected as the reference vector from the plurality of first vectors. The first pre-set between the strain cost A corresponding to the reference vector, the strain cost C corresponding to the second vector corresponding to the negative reference vector, and the strain cost B corresponding to the initial motion vector. If the relationship satisfies the first preset condition, the reference vector is used as the final motion vector of the prediction block. The first pre-set between the strain cost A corresponding to the reference vector, the strain cost C corresponding to the second vector corresponding to the negative reference vector, and the strain cost B corresponding to the initial motion vector. If the relationship does not satisfy the first preset condition and A and B satisfy the second preset condition, then the length of the first vector is to obtain the third vector. Decreased with a preset precision, the third vector is used as the first target vector of the prediction block. If A and B do not meet the second preset condition, and B and C satisfy the third preset condition, then the length of the first vector is to obtain the fourth vector. Increased with preset precision, the fourth vector is used as the first target vector of the prediction block. If B and C do not meet the third preset condition, a particular loop count is set. In each loop, the offset of the reference vector is updated first. Specifically, if the second preset relationship between A, B, C satisfies the fourth preset condition, the offset of the reference vector is updated. If the second pre-set relationship between A, B, C does not meet the fourth pre-set condition, or the loop count drops to zero, the offset of the reference vector is not updated. The loop ends. After the loop ends, the direction of the offset is determined based on the relationship between the values of A and C, and the direction of the offset includes the positive or negative direction. A fifth vector is obtained based on the reference vector, the offset, and the direction of the offset, and the fifth vector is used as the first target vector of the prediction block. Finally, the first target vector is used as the final motion vector of the prediction block. The first pre-set relationship may be a proportional relationship, a differential relationship, or a value relationship. The first pre-set condition may be that the first pre-set relationship between A, C, and B is equal to the pre-set value, or the second pre-set. The condition may be that A is equal to B, and the third pre-set condition may be that B is equal to C. 1st pre-set relationship, 1st pre-set condition, pre-set value, 2nd pre-set condition, 3rd pre-set condition, 2nd pre-set relationship , And the fourth pre-set conditions are not particularly limited in this embodiment of the invention.

For example, there are eight candidate motion vectors, and the eight candidate motion vectors are located in four directions: up, down, left, and right of the prediction block. There are two candidate motion vectors in each direction. To obtain the four first vectors, the maximum length candidate motion vector is first selected from each of the four directions. If the first vector in the upward direction corresponds to the minimum strain cost, the first vector in the upward direction is used as the reference vector, that is, the final vector of the prediction block is the positive motion vector. The first pre-set condition is that the proportional relationship between the strain cost A corresponding to the reference vector, the strain cost C corresponding to the second vector, and the strain cost B corresponding to the initial motion vector is 1. Yes, the second vector is a candidate motion vector in the opposite direction to the reference vector, and it is assumed that the preset precision is 1/4 precision. In this case, when the strain cost corresponding to the reference vector is 2, the strain cost corresponding to the first downward vector is 4, and the strain cost corresponding to the initial motion vector is 3, the reference vector. The proportional relationship between the strain cost corresponding to, the strain cost corresponding to the first downward vector, and the strain cost corresponding to the initial motion vector is (2 × B) / (A + C), and the proportional relationship is 1. Is assumed. In this case, the reference vector is used as the final motion vector of the pre-set block. If the strain cost corresponding to the first vector in the downward direction is 5 and the ratio relationship (2 × B) / (A + C) is not 1, the ratio does not satisfy the first preset condition. In this case, if A is equal to B, the length of the reference vector is reduced by a pre-set precision of 1/4 to obtain the third vector, and the third vector is the final of the pre-set block. Used as a motion vector. If A is not equal to B and B is equal to C, then the length of the reference vector is increased by a pre-set precision of 1/4 to obtain the 4th vector, and the 4th vector is preset. Used as the final motion vector of the block. If B is not equal to C, the loop count is set to 3 and the offset accuracy of the reference vector is first updated to 1/8 in each loop and a second advance between A, B, and C. It is assumed that the set relationships are K = | (AC) × 16 | and T = ((A + C) -2 × B) × 8. If K is greater than or equal to T, the fourth preset condition is satisfied and the offset accuracy of the reference vector is updated to 1/16. It is assumed that K and T are updated to K = KT and T = T / 2 in each loop. When the loop coefficient decreases to 0, the loop ends. After the loop ends, the direction of the offset is determined based on the relationship between the values of A and C. If A is greater than or equal to C, the offset direction is positive. If A is less than C, the offset direction is negative. A fifth vector is obtained based on the reference vector, the offset, and the direction of the offset, and the fifth vector is used as the first target vector of the prediction block. Finally, the first target vector is used as the final motion vector of the prediction block.

Indeed, in some embodiments, a third pre-set relationship between the strain cost corresponding to the reference vector and the strain cost corresponding to the third vector in the target direction is set in the fifth pre-set. If the above conditions are not met, the length of the first vector can be increased by the preset accuracy until the increased strain cost corresponding to the second target vector meets the second target condition. If the third preset relationship between the strain cost corresponding to the reference vector and the strain cost corresponding to the third vector in the target direction satisfies the fifth preset condition, then the first vector. The length can be reduced by the preset accuracy until the reduced strain cost corresponding to the second target vector meets the target condition. The target direction is any direction other than the direction of the reference vector. The third pre-set relationship, the fifth pre-set condition, and the target condition are not specifically limited in this embodiment of the invention.

In some embodiments, the fourth preset relationship of the strain costs corresponding to the reference vector, the strain cost corresponding to the third vector in the target direction, and the strain cost corresponding to the intermediate vector is the sixth. If the preset condition of is satisfied, the third target vector is determined by increasing the reference vector by the preset accuracy. The fourth pre-set relationship between the strain cost corresponding to the reference vector, the strain cost corresponding to the third vector in the target direction, and the strain cost corresponding to the intermediate vector is the sixth pre-set condition. If not satisfied, the third target vector is determined by decrementing the reference vector with preset precision. The intermediate vector is a motion vector located between the reference vector and the second vector. The process of increasing the reference vector with preset precision and decreasing the reference vector with preset precision has been described above and will not be described in detail here again. The fourth pre-set relationship and the sixth pre-set condition are not specifically limited in this embodiment of the invention.

FIG. 14 is a flowchart of an example of updating a motion vector during video decoding according to an embodiment of the present application. The picture block to be processed is a block to be decoded.

1401: The initial motion vector of the picture block to be processed is acquired based on the predicted motion vector of the picture block to be processed indicated by the index.

In a viable implementation, for example, in merge mode, the predicted motion vector of the picture block to be processed is used as the initial motion vector of the picture block to be processed.

The predicted motion vector of the picture block to be processed is the method shown in FIGS. 6 to 9 in the embodiment of the present application, or H. It can be acquired by any one of the existing predicted motion vector acquisition methods in the 265 standard or JEM reference mode. This is not limited. The motion vector difference can be obtained by analyzing the bitstream.

During bidirectional prediction, this step specifically: obtains the first initial motion vector of the picture block to be processed, based on the forward predicted motion vector of the picture block to be processed. , A step of acquiring a second initial motion vector of the picture block to be processed, based on the backward predicted motion vector of the picture block to be processed.

1402: Obtains a predicted block of picture blocks to be processed based on the initial motion vector and one or more preset motion vector offsets. The details are as follows.

S14021: The initial stage of the picture block to be processed from the reference frame of the picture block to be processed indicated by the reference frame index of the picture block to be processed in order to be supplied as the reference block of the picture block to be processed. Gets the picture block represented by the motion vector of.

14022: To obtain one or more final motion vectors, add the initial motion vector and one or more pre-set motion vector offsets of the picture block to be processed. Here, each final motion vector indicates a search position.

14023: Acquire one or more candidate prediction blocks at one or more search positions indicated by one or more final motion vectors. Each search position corresponds to one candidate prediction block.

14024: From one or more candidate prediction blocks, the candidate prediction block with the smallest pixel difference from the temporal prediction block is selected as the prediction block for the picture block to be processed.

During bidirectional prediction, this step specifically: From the first reference frame of the picture block to be processed, indicated by the first reference frame index of the picture block to be processed, to the picture block to be processed. Gets the first picture block indicated by the first initial motion vector of, and processes from the second reference frame of the picture block to be processed, which is indicated by the second reference frame index of the picture block to be processed. The step of getting the second picture block, which is indicated by the second initial motion vector of the picture block of interest,
A step of performing weighting processing on the first picture block and the second picture block, and a first initial motion vector and processing in order to obtain a temporally predicted block of the picture block to be processed. Add one or more pre-set motion vector offsets of the picture block of interest to get one or more first final motion vectors, the second initial motion vector and the object to be processed. A step of adding one or more pre-set motion vector offsets of a picture block to obtain one or more second final motion vectors and one or more first finals. Get one or more forward candidate prediction blocks at one or more search positions indicated by a motion vector, and one or more at one or more search positions indicated by one or more second final motion vectors. From the step of acquiring the backward candidate prediction block of and one or more forward candidate prediction blocks, the candidate prediction block having the minimum pixel difference from the temporal prediction block is selected as the forward prediction block of the picture block to be processed. Then, from one or more backward candidate prediction blocks, a step of selecting a candidate prediction block having the minimum pixel difference from the temporal prediction block as a backward prediction block of the picture block to be processed, and a picture to be processed. In order to obtain the predicted block of the block, a step of executing a weighting process on the forward predicted block and the backward predicted block is included.

The implementation of updating the motion vector will be described in detail below by using some specific embodiments. It should be understood that motion vector updates are consistent in encoders and decoders, as described in the coding method of FIG. 13 and the decoding method of FIG. Therefore, the following embodiments will be described only from the encoder or decoder. The implementation in the decoder remains consistent with that in the encoder when the description from the encoder is provided, and the implementation in the encoder is consistent with that in the decoder when the description from the decoder is provided. It should be understood that it maintains.

As shown in FIG. 15, the current decoding block is the first decoding block, and the predicted motion information of the current decoding block can be obtained. The motion vector predictors forward and backward of the current decoded block are (-10, 4) and (5, 6), respectively, and the picture order count (POC) of the current decoded block. ) POC is assumed to be 4. The POC is used to indicate the display order of the pictures, and the POCs of the reference pictures indicated by the index values of the displayed reference pictures of the pictures are 2 and 6, respectively. In this case, the POC corresponding to the current decryption block is 4, the POC corresponding to the forward predictive reference picture block is 2, and the POC corresponding to the backward predictive reference picture block is 6.

Forward and backward predictions are performed separately for the current decoding block, with the initial forward prediction block (FPB) and the initial backward decoding prediction block (FPB) of the current decoding block. backward prediction block, BPB). It is assumed that the initial forward decoding prediction block and the initial backward decoding prediction block are FPB1 and BPB1, respectively. The first decoding prediction block (DPB) of the current decoding block is obtained by weighting addition to FPB1 and BPB1 and is assumed to be DPB1.

(-10,4) and (5,6) are used as reference inputs for forward and backward motion vector predictors, with first precision motion search for forward and backward predictive reference picture blocks. Are executed separately. In this case, the first precision is 1/2 pixel precision in the 1 pixel range. The first decryption prediction block DPB1 is used as a reference. The corresponding new forward and backward decoding prediction blocks obtained by each motion search are compared with the first decoding prediction block DPB1 to obtain a new decoding prediction block with the smallest difference from DPB1 and a new decoding. The forward and backward motion vector predictors corresponding to the transformation prediction blocks are used as target motion vector predictors and are assumed to be (-11,4) and (6,6), respectively.

The target motion vector predictor is updated to (-11,4) and (6,6), and forward and backward predictions are performed separately for the first decoding block based on the target motion vector predictor. , The target decryption prediction block is acquired by weighting the acquired new forward and backward decoding prediction blocks and is assumed to be DPB2, and the decoding prediction block of the current decoding block is Updated to DPB2.

If the first precision motion search is performed on the forward predictive reference picture block and the backward predictive reference picture block, the first precision may be any specified precision, eg, an integer pixel precision. , 1/2 pixel accuracy, 1/4 pixel accuracy, or 1/8 pixel accuracy.

As shown in FIG. 16, the current decoding block is the first decoding block, and the predicted motion information of the current decoding block is acquired. The forward motion vector predictor of the current decoded block is (-21,18), the POC of the picture in which the current decoded block is located is 4, and the POC of the reference picture indicated by the index value of the reference picture is It is assumed to be 2. In this case, the POC corresponding to the current decryption block is 4 and the POC corresponding to the forward predictive reference picture block is 2.

Forward prediction is performed on the current decryption block to get the initial forward prediction block of the current decryption block. It is assumed that the initial forward decoding prediction block is FPB1. In this case, the FPB1 is used as the first decoding prediction block of the current decoding block, and the first decoding prediction block is shown as the DPB1.

(-21, 18) are used as reference inputs for the forward motion vector predictor, and a first-precision motion search is performed on the forward motion prediction reference picture block. In this case, the first precision is 1 pixel precision in the 5 pixel range. The first decryption prediction block DPB1 is used as a reference. The corresponding new forward decoding prediction block acquired by each motion search is compared with the first decoding prediction block DPB1 to obtain a new decoding prediction block with the minimum difference from DPB1 and a new decoding prediction block. The forward motion vector predictor corresponding to is used as the target motion vector predictor and is assumed to be (-19,19).

The target motion vector predictor is updated to (-19, 19), forward prediction is performed based on the target motion vector predictor, and the new forward decoding prediction block obtained is used as the target decoding prediction block. Assumed to be DPB2, the decoding prediction block of the current decoding block is updated to DPB2.

If the first precision motion search is performed on the forward predictive reference picture block and the backward predictive reference picture block, the first precision may be any specified precision, eg, an integer pixel precision. It should be noted that the accuracy may be 1/2 pixel accuracy, 1/4 pixel accuracy, or 1/8 pixel accuracy.

As shown in FIGS. 17A and 17B, the current coding block is the first coding block, and the predicted motion information of the current coding block can be obtained. The forward and backward motion vector predictors of the current coding block are (-6,12) and (8,4), respectively, and the POC of the picture in which the current decoding block is located is 8, and the index of the reference picture. It is assumed that the POCs of the reference pictures indicated by the values are 4 and 12, respectively. In this case, the POC corresponding to the current coding block is 4, the POC corresponding to the forward predictive reference picture block is 4, and the POC corresponding to the backward predictive reference picture block is 12.

The forward and backward predictions are performed separately for the current coded block to get the initial forward and backward coded predictive blocks of the current coded block. The initial forward-coded predictive block and the initial backward-coded predictive block are assumed to be FPB1 and BPB1, respectively. The first coding prediction block of the current coding block is obtained by performing weighting addition to FPB1 and BPB1 and is assumed to be DPB1.

(-6,12) and (8,4) are used as reference inputs for forward and backward motion vector predictors, with first precision motion search for forward and backward predictive reference picture blocks. Will be executed separately. The first coded prediction block DPB1 is used as a reference. The corresponding new forward and backward coded predictive blocks obtained by each motion search are compared with the first coded predictive block DPB1 to obtain a new coded predictive block with the minimum difference from the DBI and a new coded form. The forward and backward motion vector predictors corresponding to the prediction blocks are used as target motion vector predictors and are assumed to be (-11,4) and (6,6), respectively.

The target motion vector predictors are updated to (-11,4) and (6,6), forward and backward predictions are performed separately for the first coded block based on the target motion vector predictors. The target coded predictive block is obtained by performing a weighted addition on the new forward and backward coded predictive blocks obtained and is assumed to be DPB2, and the coded predictive block of the current coding block is DPB2. Will be updated to.

(-11,4) and (6,6) are then used as reference inputs for the forward and backward motion vector predictors, and the first precision motion search is the forward predictive reference picture block and the backward predictive reference picture. Executed separately for blocks. The coded prediction block DPB2 of the current coding block is used as a reference. The corresponding new forward and backward coded prediction blocks obtained in each motion search are compared with the first coded predictive block DPB2 to obtain a new coded predictive block with the smallest difference from DPB2 and a new code. The forward and backward motion vector predictors corresponding to the transformation prediction blocks are used as new target motion vector predictors and are assumed to be (-7,11) and (6,5), respectively.

The target motion vector predictors are then updated to (-7,11) and (6,5), and forward and backward predictions are based on the latest target motion vector predictors for the first coding block. Performed separately, the target coded predictive block is obtained by performing a weighted addition on the new forward and backward coded predictive blocks obtained, assumed to be DPB3, and the code of the current coding block. The conversion prediction block is updated to DPB3.

Further, the target motion vector predictor can be continuously updated according to the above method, and the number of cycles is not limited.

If the first precision motion search is performed on the forward predictive reference picture block and the backward predictive reference picture block, the first precision may be any specified precision, eg, integer pixel precision, It should be noted that it may be 1/2 pixel accuracy, 1/4 pixel accuracy, or 1/8 pixel accuracy.

It should be understood that in some feasible embodiments, the cycle ends when the conditions are met. For example, the cycle ends when the difference between DPBn and DPBn-1 is less than the threshold, where n is a positive integer greater than 2.

As shown in FIG. 18, the current decoding block is the first decoding block, and the predicted motion information of the current decoding block can be obtained. The forward and backward motion vector values of the current decoded block are (-10,4) and (5,6), respectively, and the forward and backward motion vector differences of the current decoded block are (-2,1), respectively. And (1,1), it is assumed that the POC of the picture in which the current decoding block is located is 4 and the POC of the reference picture indicated by the index value of the reference picture is 2 and 6, respectively. Therefore, the POC corresponding to the current decoded block is 4, the POC corresponding to the forward predictive reference picture block is 2, and the POC corresponding to the backward predictive reference picture block is 6.

Forward and backward predictions are performed separately for the current decoding block to get the initial forward decoding prediction block (FPB) and the initial backward decoding prediction block (BPB) of the current decoding block. do. The initial forward decoding prediction block and the initial backward decoding prediction block are assumed to be FPB1 and BPB1, respectively. The first decoding prediction block (DPB) of the current decoding block is obtained by performing weighting addition on FPB1 and BPB1 and is assumed to be DPB1.

The sum of the forward motion vector predictor and the forward motion vector difference and the sum of the backward motion vector predictor and the backward motion vector difference are (-10,4) + (-2,1) = (-12,5). And (5,6) + (1,1) = (6,7) are used as forward motion vector and backward motion vector, respectively, and the first-precision motion search refers to the forward prediction reference picture block and the backward prediction reference. Executed separately for picture blocks. In this case, the first precision is 1/4 pixel precision in 1 pixel / range. The first decryption prediction block DPB1 is used as a reference. The corresponding new forward and backward decoding prediction blocks obtained in each motion search are compared with the first decoding prediction block DPB1 to obtain a new decoding prediction block with the minimum difference from DPB1. The forward and backward motion vectors corresponding to the new decoded prediction block are used as target motion vector predictors and are assumed to be (-11,4) and (6,6), respectively.

The target motion vector is updated to (-11,4) and (6,6), forward prediction and backward prediction are executed separately for the first decoding block based on the target motion vector, and the target decoding prediction is performed. The block is acquired by performing weighting addition on the acquired new forward and backward decoding prediction blocks, is assumed to be DPB2, and the decoding prediction block of the current decoding block is updated to DPB2. ..

In order to further reflect the core technical content of the present application, the process of storing motion vectors by the encoder in the merge mode will be described below by using specific embodiments. FIG. 19 is a schematic flowchart of a method of acquiring a motion vector by an encoder according to the embodiment of the present application. The method includes the following steps.

1901: After acquiring the first processed picture block, the encoder determines one or more pre-set candidate reference blocks of the first processed picture block.

Referring to FIG. 3, the first processed picture block is a coded picture block, that is, the coded picture block is such that the partition unit of the encoder partition stores the video data. Obtained after acquisition. It should be noted that the picture block to be processed may be the current PU or the current CU. This is not specifically limited in this embodiment of the invention.

The candidate reference block includes a picture block having a preset spatial positional relationship with the first processed picture block. The preset spatial positional relationship may be the positional relationship shown in FIG. Since the candidate reference block shown in FIG. 5 is adjacent to the picture block to be processed, that is, the picture information of the candidate reference block is compared with that of the picture block to be processed. Due to their similarities, the motion vector of the candidate reference block is used as the predicted motion vector of the first processed picture block in order to predict the first processed picture block. It is possible. In some embodiments, the candidate reference block is further a virtual picture block having a pre-set spatial correlation with another real picture block or a first processed picture block. include. In this embodiment of the present application, the positional relationship between the candidate reference block and the first processed picture block is not specifically limited by a particular amount of candidate reference blocks.

1902: The encoder determines one or more motion vectors of one or more candidate reference blocks.

In this embodiment of the present application, the motion vector update process is executed for the candidate reference block. When the update process for the candidate reference block is completed, the candidate reference block has an initial motion vector and a final motion vector, that is, the candidate reference block has a corresponding initial motion vector and final motion. Memorize the vector. In some embodiments, the candidate reference block stores the initial motion vector and motion vector residuals accordingly, and the final motion vector of the candidate reference block is the initial motion vector of the candidate reference block. And can be obtained by adding the motion vector residuals.

However, if the update process for the candidate reference block is not completed, the candidate reference block has only the initial motion vector, that is, the candidate reference block correspondingly stores only the initial motion vector of the candidate reference block. ..

Since the candidate reference block has an initial motion vector and a final motion vector, the encoder uses the initial motion vector or the final motion vector of the candidate reference block as the motion vector of the candidate reference block. As a result, the motion vector of the candidate reference block is used as the predicted motion vector of the spatial candidate of the picture block to be processed.

It should be noted that for the motion vector update process in the reference block, refer to the embodiment of the present application related to FIG. It should be understood that the reference block of the embodiment related to FIG. 19 is the picture block to be processed in the embodiment related to FIG.

In step 1902, the motion vector of the candidate reference block may be determined based on the previously set range of the candidate reference block. For example, if the candidate reference block is within the preset range, the initial motion vector of the candidate reference block is used as the motion vector of the candidate reference block, or the candidate reference block is preset. If it is out of range, the final motion vector of the candidate reference block is used as the motion vector of the candidate reference block. Therefore, one or more motion vectors of one or more candidate reference blocks can be determined.

Specifically, the preset range may be the CTB row range. For example, if the CTB where the candidate reference block is located and the CTB where the first picture block to be processed is located are located on the same row, the initial motion vector of the candidate reference block is the candidate reference block. Used as a motion vector. Correspondingly, if the CTB where the candidate reference block is located and the CTB where the first picture block to be processed is located are not located in the same row, the initial motion vector of the candidate reference block is a candidate. Used as a motion vector for the reference block of. Referring to FIG. 21, the coding tree block in which the candidate reference block is located and the coding tree block in which the first picture block to be processed is located are located on different rows, and the candidate reference block is located. If the located coding tree block and the coding tree block in which the first processed picture block is located are located in different rows of adjacent spaces, then the initial motion vector of the candidate reference block is: It will be appreciated that it will be used as a motion vector for the candidate reference block. The adjacent space of different rows may be the upper space, upper left space, etc. of different rows. The specific range of spaces adjacent to different rows is not specifically limited in this embodiment of the invention.

In a related technique, the encoder may encode a plurality of picture blocks to be processed at the same time. However, when a plurality of processed picture blocks are located in the same CTB row, the plurality of processed picture blocks may be reference blocks of each other. If multiple processed picture blocks are reference blocks of each other's candidates, the updated final motion vector of another block will only be the current processed picture block after another block has been updated. It can be used as a predicted motion vector for block spatial candidates. As a result, there is a coding delay. Further, since the plurality of picture blocks to be processed are reference blocks of each other's candidates, it is necessary to wait until the other blocks are updated. As a result, the coding delay increases.

Therefore, the initial motion vector of the candidate reference block can be used directly as the motion vector of the candidate reference block, and then the motion vector of the candidate reference block is the first object to be processed. Used as a predicted motion vector for spatial candidates in a picture block. See step 1903 for this.

If the CTB where the candidate reference block is located and the CTB where the first picture block to be processed is located are located on the same row, the update process for the candidate reference block may not be completed. Therefore, the initial motion vector of the candidate reference block can be used directly as the motion vector of the candidate reference block, and there is no need to wait for the candidate reference block to be updated, and it is coded. Reduce delay. If the CTB where the candidate reference block is located and the CTB where the first process target picture block is located are not located on the same row, it is sufficient to complete the update process for the candidate reference block. Show that you have time. Therefore, the candidate reference block has a final motion vector, and the final motion vector of the candidate reference block can be used as the motion vector of the candidate reference block. Furthermore, since the final motion vector is the motion vector obtained after the initial motion vector has been refined, using the final motion vector as the motion vector of the candidate reference block guarantees the coding quality. can do.

The preset range may be the CTB block range instead. For example, when the candidate reference block and the first processed picture block are located in the same CTB, the initial motion vector of the candidate reference block is used as the motion vector of the candidate reference block. Correspondingly, if the candidate reference block and the first processed picture block are not located in the same CTB, the final motion vector of the candidate reference block is the motion vector of the candidate reference block. Used as.

If the candidate reference block and the first processed picture block are located in the same CTB, it indicates that the update process in the candidate reference block may not be completed, and the candidate reference block. The initial motion vector of is used directly as the motion vector of the candidate reference block without having to wait for the candidate reference block to be updated, reducing the coding delay. Further, if the candidate reference block and the first processed picture block are not located in the same CTB, it indicates that there is sufficient time to complete the update process for the candidate reference block. .. Further, since the final motion vector is a motion vector obtained after the initial motion vector is refined, the final of the candidate reference blocks that are not located in the same CTB as the first processed picture block. Using the motion vector as the motion vector of the candidate reference block can guarantee the coding quality.

The pre-set range may optionally be the same CTB block range or the range of left-adjacent and right-adjacent CTB blocks. For example, the CTB where the candidate reference block is located is the CTB where the first process target picture block is located, or the CTB where the candidate reference block is located is the first process target picture block. If it is a left-adjacent or right-adjacent block of the located CTB, the initial motion vector of the candidate reference block is used as the motion vector of the candidate reference block. Correspondingly, the CTB where the candidate reference block is located is not the same as the CTB where the first process target picture block is located, or the CTB where the candidate reference block is located is the first process target picture. If the block is not on the left or right side of the CTB where the block is located, the final motion vector of the candidate reference block is used as the motion vector of the candidate reference block.

If the CTB where the candidate reference block is located is the same as the CTB where the first process target picture block is located, or the CTB where the candidate reference block is located is where the first process target picture block is located. If it is a left-adjacent or right-adjacent block of the candidate CTB, it indicates that the update process for the candidate reference block may not be completed, and the initial motion vector of the candidate reference block is updated by the candidate reference block. It is used directly as the motion vector of the candidate reference block without the need to wait until it is done, reducing the coding delay. Further, the CTB where the candidate reference block is located is not the same as the CTB where the first process target picture block is located, or the CTB where the candidate reference block is located is the first process target picture block. If is not a left-adjacent or right-adjacent block of the CTB in which it is located, it indicates that there is sufficient time to complete the update process for the candidate reference block. Furthermore, since the final motion vector is the motion vector obtained after the initial motion vector has been refined, it is not possible to use the final motion vector of the candidate reference block as the motion vector of the candidate reference block. , The coding quality can be guaranteed.

1903: The encoder sets a list of predicted motion vectors of candidates based on one or more motion vectors of one or more candidate reference blocks.

See FIGS. 6-9 for how to set a list of predicted motion vectors for candidates. Details will not be explained here again.

1904: The encoder determines the reference block of the picture block to be processed. The reference block and the first processed picture block are located within the frame of the same picture.

The candidate reference block with the minimum rate distortion cost is selected from one or more candidate reference blocks of the first processable picture block as the reference block of the first processable picture block.

1905: The encoder stores the motion vector of the reference block. The motion vector of the reference block is the initial motion vector of the picture block currently being processed.

The motion vector of the reference block can be obtained from the corresponding position in the list of predicted motion vectors of the set candidates.

Steps 1901 to 1905 are processes in which the encoder acquires the initial motion vector of the first process target picture block.

1906: The encoder determines the predictive block of the first processed picture block based on the initial motion vector of the first processed picture block and one or more preset motion vector offsets. decide.

Step 1906 can be completed by using 13021 to 13024 of FIG. Details will not be explained here again.

One or more pre-set motion vector offsets can be set on a pixel-by-pixel basis. For example, the picture block represented by the initial motion vector of the picture block to be processed is used as a temporal prediction block of the picture block to be processed, and is one of the temporal prediction blocks. One or more candidate prediction blocks within pixel units or 1/2 pixel units are searched by using the temporal prediction blocks as the center. Indeed, different lengths or different pixel units may be used. One or more pre-set motion vector offsets are not specifically limited in this embodiment of the present application.

1907: After determining the motion vector of the predicted block, the encoder uses the motion vector of the predicted block as the final motion vector of the picture block to be processed.

In step 1907, the motion vector of the candidate reference block can be determined based on the pre-set range of the predicted block. For example, if the predicted block is within a preset range, the initial motion vector of the predicted block is used as the motion vector of the predicted block, and the predicted block is outside the preset range. The final motion vector of the predicted block is used as the motion vector of the predicted block.

Specifically, the preset range may be the CTB row range. For example, if the CTB where the prediction block is located and the CTB where the first picture block to be processed is located are located on the same row, the initial motion vector of the prediction block is used as the motion vector of the prediction block. Will be done. Correspondingly, if the CTB where the prediction block is located and the CTB where the first picture block to be processed is located are not located in the same row, the initial motion vector of the prediction block is the motion vector of the prediction block. Used as. Referring to FIG. 21, the coding tree block in which the prediction block is located and the coding tree block in which the first processing target picture block is located are located on different rows, and the prediction block is located in the coding tree block. If the tree block and the coding tree block in which the first processed picture block is located are located in adjacent spaces of different rows, the initial motion vector of the predicted block is the motion of the predicted block. It will be understood that it is used as a vector. The adjacent space of different rows may be the upper space, upper left space, etc. of different rows. The particular range of adjacent spaces in different rows is not specifically limited in this embodiment of the invention.

If the CTB where the prediction block is located and the CTB where the picture block to be processed first is located are located on the same row, the update processing for the prediction block may not be completed. Therefore, the initial motion vector of the prediction block can be used directly as the motion vector of the prediction block, and it is not necessary to wait until the prediction block is updated, which reduces the coding delay. If the CTB where the prediction block is located and the CTB where the first processing target picture block is located are not located on the same row, it has enough time to complete the update process for the prediction block. Show that. Therefore, the predicted block has a final motion vector, and the final motion vector of the predicted block can be used as a motion vector of the candidate reference block. Furthermore, since the final motion vector is the motion vector obtained after the initial motion vector has been refined, using the final motion vector as the motion vector of the prediction block guarantees the coding quality. Can be done.

In some embodiments, the preset range may be the CTB block range instead. In some other embodiments, the pre-set range may be the same CTB block instead. Determining the motion vector of the predictive block is similar to determining the motion vector of the candidate reference block. See step 1902 for this. Details will not be explained here again.

1908: The encoder updates the initial motion vector of the first processed picture block in the candidate list to the final motion vector of the first processed picture block.

1909: The encoder stores the motion vector residual of the picture block to be processed in the target storage space.

The motion vector residual is the motion vector difference between the final motion vector of the first processed picture block and the initial motion vector of the first processed picture block.

The target storage space is a space for storing the motion vector residuals by the encoder. For example, the target storage space may be the reference picture storage storage 64 of FIG.

It should be noted that in merge mode in related technology, the target storage space stores zero vectors or the target storage space does not store data. However, in the merge mode of this embodiment of the present application, the motion vector residuals of the picture block to be processed are stored in the target storage space, and as a result, the target storage space can be completely used. Not only that, it is also possible to flexibly select the motion vector of the other picture block when the motion vector of the other picture block is required for the first processed picture block. .. As shown in steps 1902 and 1907, the coding efficiency of the encoder is further improved.

Further, in some implementations, in step 1909, the final motion vector of the first processed picture block is instead used as the motion vector residual of the first processed picture block. , That is, the final motion vector of the picture block to be processed is stored in the target storage space, that is, the picture block to be processed first. The motion vector residuals of are not stored in the target storage space. In this case, if the final motion vector of the picture block to be processed is required for another block, the final motion vector of the picture block to be processed is the first. Can be obtained directly from the encoder's target storage space without having to add the initial motion vector of the picture block to be processed and the motion vector residuals of the first picture block to be processed. It is possible.

See FIG. 21 for the specific storage method of step 1909.

It should be noted that there are no strict requirements for the sequences of steps 1908 and 1909, and the encoder may optionally perform step 1909 prior to step 1908. This is not specifically limited in this embodiment of the present application.

Steps 1906 to 1909 are processes in which the encoder acquires the final motion vector of the first process target picture block.

1910: The encoder encodes one or more first identification information corresponding to the initial motion vector of the picture block to be processed or the initial motion vector of the picture block to be processed in the candidate list into a bitstream. To become.

The identification information may include an index. Indeed, in some implementations, the identification information may further include a predictive mode. The index is used to indicate the corresponding motion vector in the list of candidate predicted motion vectors, which is the initial motion vector or final motion vector of the first processed picture block. , Prediction modes include merge mode, AMVPA mode, and skip mode. In this embodiment of the present application, the prediction mode is the merge mode. Indeed, in another embodiment, the prediction mode may be another mode. This is not limited to this embodiment of the present application.

In the above implementation, when encoding the current coding block, the encoder replaces the final motion vector of the reference block with the unupdated initial motion vector of the reference block to the current coding block. Used as the predicted motion vector of. In this way, if the motion vector of the reference block is needed for the current coding block, the relevant steps need to wait until the final motion vector of the reference block is updated. It can be performed without, while compensating for a reduction in coding delay, while updating the motion vector to improve coding efficiency.

When the inter-prediction mode used by the encoder is forward or backward prediction, the motion vector of the candidate reference block, the motion vector of the reference block, and the motion vector of the predictive block shall include the forward or backward motion vector. Should be noted. Correspondingly, when a list of candidate predicted motion vectors is set, the list of candidate predicted motion vectors contains only list 0 or list 1 and is the initial motion vector of the picture block to be processed. Contains an early forward motion vector or an early backward motion vector, and the final motion vector of the picture block to be processed includes a final forward motion vector or a final backward motion vector.

When the inter-prediction mode used by the encoder is bidirectional prediction, the motion vector of the candidate reference block, the motion vector of the reference block, and the motion vector of the prediction block include the first motion vector and the second motion vector. , The first motion vector is a forward motion vector, and the second motion vector is a backward motion vector. Correspondingly, when a list of candidate predicted motion vectors is set, the list of candidate predicted motion vectors includes list 0 and list 1, and the initial motion vector of the picture block to be processed is , A first initial motion vector and a second initial motion vector, and the final motion vector of the picture block to be processed is the first final motion vector and the second final motion vector. including. The first initial motion vector and the first final motion vector are forward motion vectors, and the second initial motion vector and the second final motion vector are backward motion vectors.

The above content describes the encoder. Correspondingly, the process of storing motion vectors by the decoder in the merge mode will be described below using specific embodiments. FIG. 20 is a schematic flowchart of a method of acquiring a motion vector by a decoder according to the embodiment of the present application. The method includes the following steps.

2001: After receiving the bitstream corresponding to the picture block to be processed, the decoder analyzes the bitstream to obtain the second identification information and the third identification information. Here, the second discriminant information is used to determine the initial predicted motion vector of the picture block to be processed, and the third discriminant information is the motion vector residuals analyzed during decoding. Used to indicate if it is necessary.

The second identification information may include an index used to determine the initial motion vector of the picture block to be processed. Referring to steps 2002 and 2004, in some embodiments, the second identification information may further include a predictive mode type.

The third identification information is used to indicate whether the motion vector residuals need to be analyzed during decoding. In this embodiment of the present application, the motion vector residuals do not need to be analyzed in merge mode, but in another embodiment the motion vector residuals may be analyzed. For example, in the SMVP mode, when the motion vector residual of the picture block to be processed in the second process is added to the bitstream according to the instruction of the third identification information, the decoder analyzes the bitstream to obtain the motion vector residual. Can be obtained. In this case, the decoder stores the motion vector residuals directly in the target storage space.

2002: After receiving the bitstream corresponding to the second processed picture block, the decoder has one or more candidate reference blocks and one or more candidate for the second processed picture block. Determine with one or more motion vectors of the reference block.

The reference block of one or more candidates includes a picture block having a spatial positional relationship set in advance with the picture block to be processed. See FIG. 5 for determining candidate reference blocks. Details will not be explained here again.

In this embodiment of the present application, motion vector update processing is performed on the candidate reference block. When the update process for the candidate reference block is completed, the candidate reference block has an initial motion vector and a final motion vector, that is, the candidate reference block has an initial motion vector and a final motion vector. And remember. In some embodiments, the candidate reference block stores the initial motion vector and motion vector residuals, and the final motion vector of the candidate reference block is the initial motion vector and motion of the candidate reference block. It can be obtained by adding a vector residual.

However, if the update process of the candidate reference block is not completed, the candidate reference block has only the initial motion vector, that is, the candidate reference block stores only the initial motion vector of the candidate reference block. do.

Since the candidate reference block has an initial motion vector and a final motion vector, the decoder uses the initial motion vector or the final motion vector of the candidate reference block as the motion vector of the candidate reference block. As a result, the motion vector of the candidate reference block is used as the predicted motion vector of the spatial candidate of the second processed picture block.

It should be noted that for the motion vector update process for the reference block and the acquisition of the initial motion vector, see embodiments of the present application related to FIG. It should be understood that the reference block of the embodiment related to FIG. 20 is the picture block to be processed according to the embodiment of FIG.

In step 1802, the motion vector of the candidate reference block may be determined based on the previously set range of the candidate reference block. For example, if the candidate reference block is within the preset range, the initial motion vector of the candidate reference block is used as the motion vector of the candidate reference block and the candidate reference block is preset. If it is out of range, the final motion vector of the candidate reference block is used as the motion vector of the candidate reference block. Therefore, one or more motion vectors of one or more candidate reference blocks can be determined.

Specifically, the preset range may be the CTB row range. For example, if the CTB where the candidate reference block is located and the CTB where the second picture block to be processed is located are located on the same row, the initial motion vector of the candidate reference block is the candidate reference block. Used as a motion vector. Correspondingly, if the CTB where the candidate reference block is located and the CTB where the second picture block to be processed is located are not located on the same row, the initial motion vector of the candidate reference block is a candidate. Used as a motion vector for the reference block of. Referring to FIG. 21, the coding tree block in which the candidate reference block is located and the coding tree block in which the second picture block to be processed is located are located on different rows, and the candidate reference block is located. If the coding tree block in which it is located and the coding tree block in which the second picture block to be processed is located are located in adjacent spaces on different rows, the initial motion vector of the candidate reference block is It will be appreciated that it is used as a motion vector for candidate reference blocks. The adjacent space of different rows may be the upper space, upper left space, etc. of different rows. The specific range of spaces adjacent to different rows is not specifically limited in this embodiment of the invention.

In a related technique, the decoder may simultaneously decode a plurality of picture blocks to be processed. However, when a plurality of processed picture blocks are located in the same CTB row, the plurality of processed picture blocks may be reference blocks of each other. If multiple processed picture blocks are reference blocks of each other's candidates, the updated final motion vector of another block will only be the current processed picture block after another block has been updated. It can be used as a predicted motion vector for block spatial candidates. As a result, there is a decryption delay. Further, since the plurality of picture blocks to be processed are reference blocks of each other's candidates, it is necessary to wait until the other blocks are updated. As a result, the decryption delay increases.

Therefore, the initial motion vector of the candidate reference block can be used directly as the motion vector of the candidate reference block, and then the motion vector of the candidate reference block is the second object to be processed. Used as a predicted motion vector for spatial candidates in a picture block. See step 1803 for this.

If the CTB where the candidate reference block is located and the CTB where the second picture block to be processed is located are located on the same row, the update process for the candidate reference block may not be completed. Therefore, the initial motion vector of the candidate reference block can be used directly as the motion vector of the candidate reference block, without having to wait for the candidate reference block to be updated, and decoding. Reduce delay. If the CTB where the candidate reference block is located and the CTB where the second processed picture block is located are not located on the same row, it is sufficient to complete the update process for the candidate reference block. Show that you have time. Therefore, the candidate reference block has a final motion vector, and the final motion vector of the candidate reference block can be used as the motion vector of the candidate reference block. Furthermore, since the final motion vector is the motion vector obtained after the initial motion vector has been refined, using the final motion vector as the motion vector of the candidate reference block guarantees decoding quality. can do.

The preset range may be the CTB block range instead. For example, when the candidate reference block and the second processed picture block are located in the same CTB, the initial motion vector of the candidate reference block is used as the motion vector of the candidate reference block. Correspondingly, if the candidate reference block and the second processed picture block are not located in the same CTB, the final motion vector of the candidate reference block is the motion vector of the candidate reference block. Used as.

If the candidate reference block and the second processed picture block are located in the same CTB, it indicates that the update process in the candidate reference block may not be completed, and the candidate reference block. The initial motion vector of is used directly as the motion vector of the candidate reference block without having to wait for the candidate reference block to be updated, reducing the decoding delay. Further, if the candidate reference block and the second processed picture block are not located in the same CTB, it indicates that there is sufficient time to complete the update process for the candidate reference block. .. Further, since the final motion vector is a motion vector obtained after the initial motion vector is refined, the final of the candidate reference blocks that are not located in the same CTB as the second processed picture block. Using the motion vector as the motion vector of the candidate reference block can guarantee the decoding quality.

The preset range may optionally be the same CTB block range, or the range of left-adjacent and right-adjacent CTB blocks. For example, the CTB where the candidate reference block is located is the CTB where the second process target picture block is located, or the CTB where the candidate reference block is located is the second process target picture block. If it is a left-adjacent or right-adjacent block of the located CTB, the initial motion vector of the candidate reference block is used as the motion vector of the candidate reference block. Correspondingly, the CTB where the candidate reference block is located is not the same as the CTB where the second process target picture block is located, or the CTB where the candidate reference block is located is the second process target picture. If the block is not on the left or right side of the CTB where the block is located, the final motion vector of the candidate reference block is used as the motion vector of the candidate reference block.

If the CTB where the candidate reference block is located is the same as the CTB where the second process target picture block is located, or the CTB where the candidate reference block is located is the second process target picture block. If it is a left-adjacent or right-adjacent block of the located CTB, it indicates that the update process for the candidate reference block may not be completed, and the initial motion vector of the candidate reference block is that the candidate reference block It is used directly as the motion vector of the candidate reference block without having to wait for it to be updated, reducing the decoding delay. Further, the CTB where the candidate reference block is located is not the same as the CTB where the second process target picture block is located, or the CTB where the candidate reference block is located is the second process target picture block. If is not a left-adjacent or right-adjacent block of the CTB in which it is located, it indicates that there is sufficient time to complete the update process for the candidate reference block. Furthermore, since the final motion vector is the motion vector obtained after the initial motion vector has been refined, it is not possible to use the final motion vector of the candidate reference block as the motion vector of the candidate reference block. , Decryption quality can be guaranteed.

2003: The decoder sets a list of predicted motion vectors of candidates based on one or more motion vectors of one or more candidate reference blocks.

Step 2003 is the same as step 1903, i.e., the way the decoder sets the list of candidate predicted motion vectors is consistent with the way the encoder sets the list of candidate predicted motion vectors. , The list of predicted motion vectors of the set candidates is the same. The encoder is an encoder that encodes the bitstream received by the decoder.

2004: The decoder determines the reference block of the picture block to be processed and the motion vector of the reference block from the list of predicted motion vectors of the candidates based on the second identification information.

The motion vector indicated by the index of the second identification information is selected from the list of candidate predicted motion vectors based on the index. The motion vector is the initial motion vector of the second processed picture block, and the motion vector is also the motion vector of the reference block of the second processed picture block.

2005: The encoder stores the motion vector of the reference block as the initial motion vector of the second processed picture block.

Steps 2001 to 2005 are processes in which the decoder acquires the initial motion vector of the picture block to be processed.

2006: The decoder determines the predictive block of the second processed picture block based on the initial motion vector of the second processed picture block and one or more preset motion vector offsets. decide.

Step 2006 can be completed by using 1401 to 1404 of FIG. Details will not be explained here again.

See step 1906 for one or more preset motion vector offsets. Details will not be explained here again.

It should be noted that there are unspecified sequences that perform steps 2005 and 2006, and step 2006 may be performed before step 2005 instead. The sequences that perform steps 2005 and 2006 are not specifically limited in this embodiment of the present application.

2007: After determining the motion vector of the predicted block, the decoder uses the motion vector of the predicted block as the final motion vector of the picture block to be processed.

In step 2007, the motion vector of the candidate reference block can be determined based on the pre-set range of the predicted block. For example, if the predicted block is within the preset range, the initial motion vector of the predicted block is used as the motion vector of the predicted block, and if the predicted block is outside the preset range, the prediction is made. The final motion vector of the block is used as the motion vector of the predicted block.

Specifically, the preset range may be the CTB row range. For example, if the CTB where the prediction block is located and the CTB where the second picture block to be processed is located are located on the same row, the initial motion vector of the prediction block is used as the motion vector of the prediction block. Will be done. Correspondingly, if the CTB where the prediction block is located and the CTB where the second picture block to be processed is located are not located in the same row, the initial motion vector of the prediction block is the motion vector of the prediction block. Used as. Referring to FIG. 21, the coding tree block in which the prediction block is located and the coding tree block in which the second processing target picture block is located are located on different rows, and the prediction block is located in the coding tree block. If the tree block and the coding tree block where the second processed picture block is located are located in adjacent spaces on different rows, the initial motion vector of the predictor block is the motion of the predictor block. It will be understood that it is used as a vector. The adjacent space of different rows may be the upper space, upper left space, etc. of different rows. The particular range of adjacent spaces in different rows is not specifically limited in this embodiment of the invention.

If the CTB where the prediction block is located and the CTB where the second picture block to be processed is located are located on the same row, the update processing for the prediction block may not be completed. Therefore, the initial motion vector of the predicted block can be used directly as the motion vector of the predicted block, and it is not necessary to wait until the predicted block is updated, which reduces the decoding delay. If the CTB where the predicted block is located and the CTB where the second processed picture block is located are not located on the same row, it has enough time to complete the update process for the predicted block. Show that. Therefore, the predicted block has a final motion vector, and the final motion vector of the predicted block can be used as a motion vector of the candidate reference block. Furthermore, since the final motion vector is the motion vector obtained after the initial motion vector has been refined, using the final motion vector as the motion vector of the prediction block ensures decoding quality. Can be done.

In some embodiments, the preset range may be the CTB block range instead. In some other embodiments, the preset range may be the same CTB block range instead. Determining the motion vector of the predictive block is similar to determining the motion vector of the candidate reference block. See step 1802 for this. Details will not be explained here again.

Steps 2006 and 2007 are processes in which the decoder acquires the final motion vector of the picture block to be processed.

2008: The motion vector residual of the picture block to be processed second is stored in the target storage space.

The target storage space is a space for storing motion vector residuals by the decoder. For example, the target storage space may be the reference picture storage 92 of FIG.

It should be noted that in merge mode in related technology, the target storage space stores zero vectors or the target storage space does not store data. However, in the merge mode of this embodiment of the present application, the motion vector residuals of the picture block to be processed are stored in the target storage space, and as a result, the target storage space can be completely used. Not only that, it is also possible to flexibly select the motion vector of the other picture block when the motion vector of the other picture block is required for the second processed picture block. .. As shown in steps 2002 and 2007, the decoding efficiency of the decoder is further improved.

Further, in some implementations, in step 2008, the final motion vector of the second processed picture block is instead used as the motion vector residual of the second processed picture block. , That is, the final motion vector of the second processed picture block is stored in the target storage space and is stored in the target storage space, that is, the second processed picture block. The motion vector residuals of are not stored in the target storage space. In this case, if the final motion vector of the second processed picture block is required for another block, the final motion vector of the second processed picture block is the second. Can be obtained directly from the decoder's target storage space without the need to add the initial motion vector of the picture block to be processed and the motion vector residuals of the second picture block to be processed. It is possible.

See FIG. 21 for a specific storage method in this step 2008.

It should be noted that there are no strict requirements for the sequence between steps 2007 and 2008, and the decoder may optionally perform step 2007 before step 2008. This is not specifically limited in this embodiment of the present application.

2009: The decoder acquires the picture information corresponding to the second processed picture block based on the motion vector residual of the second processed picture block and the initial motion difference.

The picture information includes pixel information used to identify the original picture of the second processable picture block.

The decoder acquires the picture information corresponding to the picture block to be processed second based on the motion vector residual of the picture block to be processed and the initial motion difference. See FIG. 3 for this. Details will not be explained here again.

Accordingly, the decoder may instead acquire the picture information corresponding to the second processed picture block based on the final motion vector of the second processed picture block. ..

In the above implementation, when decoding the current decrypted block, the decoder updates the reference block as the predicted motion vector of the current decoded block instead of the final motion vector of the reference block. Use an early motion vector that is not. Thus, if the motion vector of the reference block is required for the current decoded block, the relevant steps are performed without having to wait for the final motion vector of the reference block to be updated. It is possible to ensure that the decoding delay is reduced, while the decoding efficiency is improved as the motion vector is updated.

To further reflect the motion vector storage method, specific embodiments are used herein to illustrate the storage steps. The specific explanation is as follows:

In the merge mode, there may be two MVs in each prediction direction, the first MV information and the second MV information, respectively. For example, the first MV and the second MV may be the initial motion vector of the reference block and the final motion vector of the reference block, respectively. The MVD derivation process is introduced into merge mode. For example, in the forward prediction process, if the first MV is MV0 and the second MV is MV0', then there is a derivative relationship: MV0'= MV0 + (-) MVD0. Two MVs, eg, the initial motion vector. By storing the index value of and the previous MVD, i.e., the difference between the two MVs, the first MV can be used under certain conditions and the second MV is used under certain conditions. It is possible to avoid delays.

See step 1902 or 2002 for specific conditions. Details will not be explained here again.

In a feasible implementation
In non-merged mode, the target storage space is used to store the motion vector residuals of the reference block, and the motion vector residuals of the reference block are predicted for the final motion vector of the reference block and the reference block. It is the difference from the motion vector.
In merge mode, the target storage space is used to store the selected final motion vector, or the target storage space is the initial motion vector of the reference block and the selected final motion vector. Used to store the difference to and from the motion vector.

Another embodiment of the present application is:

As shown in FIG. 22, the current coding block is the first coding block, which provides the predicted motion information of the current coding block, the reference block is in merge mode, and the motion information of the reference block. The position of is not in the same CTB line range as the current coding block. In this case, the forward motion vector predictor (-21,18) of the current coding block is obtained by adding the forward motion vector (-22,18) and the forward MVD (1,0) of the reference block. The backward motion vector predictor (1,12) of the current coding block is obtained by adding the backward motion vector (2,12) and the backward MVD (-1,0) of the reference block and is the current coding block. The POC of the picture in which the block is located is 4, the POC of the forward reference picture indicated by the index value of the reference picture is 2, and the POC of the backward reference picture indicated by the index value of the reference picture is 6. In this case, the POC corresponding to the current coding block is 4, the POC corresponding to the forward predictive reference picture block is 2, and the POC corresponding to the backward predictive reference picture block is 6.

(-21, 18) are used as a reference input for the forward motion vector predictor, and a first-precision motion search is performed on the forward predictive reference picture block. In this case, the first precision is one pixel precision in the one pixel range. The first forward prediction block is used as a reference and the corresponding new forward prediction block is acquired by each motion search. (1, 12) is used as a reference input for the backward motion vector predictor, and a first-precision motion search is performed on the backward predictive reference picture block. In this case, the first precision is one pixel precision in the one pixel range. The first backward prediction block is used as a reference and the corresponding new backward prediction block is acquired by each motion search. The new forward prediction block is compared with the new backward prediction block to get the forward prediction block with the minimum difference and the backward prediction block with the minimum difference, and into the forward motion vector predictor and the backward prediction block corresponding to the forward prediction block. The corresponding backward motion vector predictors are used as target motion vector predictors and are assumed to be (-21,19) and (1,11), respectively.

The forward MVD (0,1) of the current coding block is obtained by subtracting the forward motion vector predictor (-21,18) from the forward target motion vector predictor (-21,19) and is the current The backward MVD (0, -1) of the coding block is obtained by subtracting the backward motion vector predictor (1,12) from the backward target motion vector predictor (1,11). The MVD calculation method 1 is as follows: MVD0 = (-21,19) − (-21,18) and MVD1 = (1,11) − (1,12); or MVD0 = (-21,19). − (-21, 18) and MVD1 = −MVD0.

Further, forward and backward predictions are performed separately for the first coding block based on the target motion vector predictor, and the obtained forward and backward prediction blocks are used as the target decoding prediction block. The prediction block of the current coding block is updated.

If the first precision motion search is performed on the forward predictive reference picture block and the backward predictive reference picture block, the first precision is any specified precision, eg, integer pixel precision, 1/2. It may be pixel precision, 1/4 pixel precision, or 1/8 pixel precision.

Another embodiment of the present application is:

As shown in FIG. 22, the current decoding block is the first decoding block, which provides the predicted motion information of the current coding block, the reference block is in merge mode, and the motion information of the reference block. The position of is not in the same CTB range as the current coding block. In this case, the forward motion vector predictor (-21,18) of the current coding block is obtained by adding the forward motion vector (-22,18) and the forward MVD (1,0) of the reference block. The backward motion vector predictor (1,12) of the current coding block is obtained by adding the backward motion vector (2,12) and the backward MVD (-1,0) of the reference block and is the current coding block. The POC of the picture in which the block is located is 4, the POC of the forward reference picture indicated by the index value of the reference picture is 2, and the POC of the backward reference picture indicated by the index value of the reference picture is 6. In this case, the POC corresponding to the current coding block is 4, the POC corresponding to the forward predictive reference picture block is 2, and the POC corresponding to the backward predictive reference picture block is 6.

(-21, 18) are used as a reference input for the forward motion vector predictor, and a first-precision motion search is performed on the forward predictive reference picture block. In this case, the first precision is one pixel precision in the one pixel range. The first forward prediction block is used as a reference and the corresponding new forward prediction block is acquired by each motion search. (1, 12) is used as a reference input for the backward motion vector predictor, and a first-precision motion search is performed on the backward predictive reference picture block. In this case, the first precision is one pixel precision in the one pixel range. The first backward prediction block is used as a reference and the corresponding new backward prediction block is acquired by each motion search.
The new forward prediction block is compared with the new backward prediction block to get the forward prediction block with the minimum difference and the backward prediction block with the minimum difference, and into the forward motion vector predictor and the backward prediction block corresponding to the forward prediction block. The corresponding backward motion vector predictors are used as target motion vector predictors and are assumed to be (-21,19) and (1,11), respectively.

The forward MVD (0,1) of the current coding block is obtained by subtracting the forward motion vector predictor (-21,18) from the forward target motion vector predictor (-21,19) and is the current The backward MVD (0, -1) of the coding block is obtained by subtracting the backward motion vector predictor (1,12) from the backward target motion vector predictor (1,11). The MVD calculation method 1 is as follows: MVD0 = (-21,19) − (-21,18) and MVD1 = (1,11) − (1,12); or MVD0 = (-21,19). − (-21, 18) and MVD1 = −MVD0.

Further, forward and backward predictions are performed separately for the first coding block based on the target motion vector predictor, and the obtained forward and backward prediction blocks are used as the target decoding prediction block. The prediction block of the current coding block is updated.

If the first precision motion search is performed on the forward predictive reference picture block and the backward predictive reference picture block, the first precision is any specified precision, eg, integer pixel precision, 1/2. It may be pixel precision, 1/4 pixel precision, or 1/8 pixel precision.

Another embodiment of the present application is:

As shown in FIG. 22, the current coding block is the first coding block, which provides the predicted motion information of the current coding block, the reference block is in merge mode, and the motion information of the reference block. The position of is not in the same CTB line range as the current coding block.
In this case, the forward motion vector predictor (-21,18) of the current coding block is obtained by adding the forward motion vector (-22,18) and the forward MVD (1,0) of the reference block. The backward motion vector predictor (1,12) of the current coding block is obtained by adding the backward motion vector (2,12) and the backward MVD (-1,0) of the reference block and is the current coding block. The POC of the picture in which the block is located is 4, the POC of the forward reference picture indicated by the index value of the reference picture is 2, and the POC of the backward reference picture indicated by the index value of the reference picture is 6. In this case, the POC corresponding to the current coding block is 4, the POC corresponding to the forward predictive reference picture block is 2, and the POC corresponding to the backward predictive reference picture block is 6.

(-21, 18) are used as a reference input for the forward motion vector predictor, and a first-precision motion search is performed on the forward predictive reference picture block. In this case, the first precision is one pixel precision in the one pixel range. The first forward prediction block is used as a reference and the corresponding new forward prediction block is acquired by each motion search. (1, 12) is used as a reference input for the backward motion vector predictor, and a first-precision motion search is performed on the backward predictive reference picture block. In this case, the first precision is one pixel precision in the one pixel range. The first backward prediction block is used as a reference and the corresponding new backward prediction block is acquired by each motion search. The new forward prediction block is compared with the new backward prediction block to get the forward prediction block with the minimum difference and the backward prediction block with the minimum difference, and into the forward motion vector predictor and the backward prediction block corresponding to the forward prediction block. The corresponding backward motion vector predictors are used as target motion vector predictors and are assumed to be (-21,19) and (1,11), respectively.

(-21,19) is then used as a reference input for the forward motion vector predictor, and a first precision motion search is performed on the forward predictive reference picture block. In this case, the first precision is 1/2 pixel precision. The first forward prediction block is used as a reference and the corresponding new forward prediction block is acquired by each motion search. (1, 11) is used as a reference input for the backward motion vector predictor, and a first-precision motion search is performed on the backward predictive reference picture block. In this case, the first precision is 1/2 pixel precision. The first backward prediction block is used as a reference and the corresponding new backward prediction block is acquired by each motion search. The new forward prediction block is compared with the new backward prediction block to get the forward prediction block with the minimum difference and the backward prediction block with the minimum difference.
The forward motion vector predictor corresponding to the forward prediction block and the backward motion vector predictor corresponding to the backward prediction block are used as the target motion vector predictor.

The forward MVD (0,1) of the current coding block is obtained by subtracting the forward motion vector predictor (-21,18) from the forward target motion vector predictor (-21,19) and is the current The backward MVD (0, -1) of the coding block is obtained by subtracting the backward motion vector predictor (1,12) from the backward target motion vector predictor (1,11). The MVD calculation method 1 is as follows: MVD0 = (-21,19) − (-21,18) and MVD1 = (1,11) − (1,12); or MVD0 = (-21,19). − (-21, 18) and MVD1 = −MVD0.

Further, forward and backward predictions are performed separately for the first coding block based on the target motion vector predictor, and the obtained forward and backward prediction blocks are used as the target decoding prediction block. The prediction block of the current coding block is updated.

If the first precision motion search is performed on the forward predictive reference picture block and the backward predictive reference picture block, the first precision is any specified precision, eg, integer pixel precision, 1/2. It may be pixel precision, 1/4 pixel precision, or 1/8 pixel precision.

Another embodiment of the present application is:

The current decoding block is the first coding block, and the predicted motion information of the current coding block is acquired. In this case, the forward motion vector predictor (-21,18) of the current coding block is the forward motion vector (-21,18) of the reference block and the backward motion vector predictor (1) of the current coding block. , 12) are the backward movement vectors (1,12) of the reference block, the POC of the picture in which the current coding block is located is 4, and the POC of the forward reference picture indicated by the index value of the reference picture is 2. The POC of the back-referenced picture indicated by the index value of the reference picture is 6. In this case, the POC corresponding to the current coding block is 4, the POC corresponding to the forward predictive reference picture block is 2, and the POC corresponding to the backward predictive reference picture block is 6.

(-21, 18) are used as a reference input for the forward motion vector predictor, and a first-precision motion search is performed on the forward predictive reference picture block. In this case, the first precision is one pixel precision in the one pixel range. The first forward prediction block is used as a reference and the corresponding new forward prediction block is acquired by each motion search. (1, 12) is used as a reference input for the backward motion vector predictor, and a first-precision motion search is performed on the backward predictive reference picture block. In this case, the first precision is one pixel precision in the one pixel range. The first backward prediction block is used as a reference and the corresponding new backward prediction block is acquired by each motion search. The new forward prediction block is compared with the new backward prediction block to get the forward prediction block with the minimum difference and the backward prediction block with the minimum difference, and into the forward motion vector predictor and the backward prediction block corresponding to the forward prediction block. The corresponding backward motion vector predictors are used as target motion vector predictors and are assumed to be (-21,19) and (1,11), respectively.

The forward MVD (0,1) of the current coding block is obtained by subtracting the forward motion vector predictor (-21,18) from the forward target motion vector predictor (-21,19) and is the current The backward MVD (0, -1) of the coding block is obtained by subtracting the backward motion vector predictor (1,12) from the backward target motion vector predictor (1,11). The MVD calculation method 1 is as follows: MVD0 = (-21,19) − (-21,18) and MVD1 = (1,11) − (1,12); or MVD0 = (-21,19). − (-21, 18) and MVD1 = −MVD0.

Further, forward and backward predictions are performed separately for the first coding block based on the target motion vector predictor, and the obtained forward and backward prediction blocks are used as the target decoding prediction block. The prediction block of the current coding block is updated.

If the first precision motion search is performed on the forward predictive reference picture block and the backward predictive reference picture block, the first precision is any specified precision, eg, integer pixel precision, 1/2. It may be pixel precision, 1/4 pixel precision, or 1/8 pixel precision.

In some embodiments, the picture block to be processed may contain a plurality of sub-blocks, and the codec is the final of the picture blocks to be processed based on the final motion vector of each sub-block. Motion vector can be obtained. In a possible implementation, each sub-block has an initial motion vector and a plurality of pre-set offset vectors of the sub-block. The codec has an initial motion vector of any one of a plurality of sub-blocks and a plurality of pre-set offset vectors in order to obtain the final motion vector of multiple candidates of any sub-block. Can be added separately. The codec uses, as the final motion vector of any sub-block, the candidate final motion vector corresponding to the minimum distortion cost in the final motion vector of multiple candidates of any sub-block. Similarly, the codec can obtain the final motion vector of each sub-block, and then the codec can obtain the final motion vector of the picture block to be processed based on the final motion vector of each sub-block. Motion vector can be obtained. The process of obtaining the final motion vector of the picture block to be processed based on the final motion vector of each sub-block is not specifically limited in this embodiment of the present invention.

Before obtaining the final motion vector of multiple candidates for any subblock, the codec must first determine the initial motion vector of any subblock and the preset offset vectors. It should be noted that there is. The method of determining the initial motion vector of any sub-block is similar to the method of determining the initial motion vector of the picture block to be processed and is not described in this embodiment of the invention. Furthermore, the method of determining multiple pre-set offset vectors of any sub-block is similar to the method of determining multiple pre-set offset vectors of the picture block to be processed. Not described in this embodiment of the invention.

FIG. 23 is a schematic block diagram of the motion vector acquisition device 2300 according to the embodiment of the present application. Device 2300 is:
A decision module 2301 configured to determine a reference block for a picture block to be processed, wherein the reference block and the picture block to be processed have a preset temporal or spatial correlation and are referenced. The block has an initial motion vector and one or more preset motion vector offsets, and the initial motion vector of the reference block is obtained based on the predicted motion vector of the reference block and is a reference block. The prediction block of is obtained based on the initial motion vector and one or more pre-set motion vector offsets, with a decision module,
Includes acquisition module 2302 configured to use the initial motion vector of the reference block as the predicted motion vector of the picture block to be processed.

In a feasible implementation, the acquisition module 2302 is configured to use the predicted motion vector of the reference block as the initial motion vector of the reference block, or the predicted motion vector of the reference block and the reference block. It is further configured to obtain the initial motion vector of the reference block by adding the motion vector difference of.

In a feasible implementation, acquisition module 2302 further acquires the picture block represented by the initial motion vector of the reference block from the reference frame of the reference block so that it acts as a temporal prediction block of the reference block. To get one or more actual motion vectors, add the initial motion vector of the reference vector and one or more preset motion vector offsets (where each actual motion vector is Obtain one or more candidate prediction blocks at one or more search positions indicated by one or more actual motion vectors (indicating one or more search positions) (where each search position is one). A candidate prediction block having the smallest pixel difference from the temporal prediction block (corresponding to the candidate prediction block) is configured to be selected from one or more candidate prediction blocks as the reference block prediction block. ..

In a feasible implementation, the device 2300 is used for bidirectional prediction, the reference frame includes a reference frame in the first direction and a reference frame in the second direction, and the initial motion vector is the initial motion vector in the first direction. It contains a motion vector and an initial motion vector in the second direction, and the acquisition module 2302 is specifically indicated by the initial motion vector in the first direction of the reference block from the reference frame in the first direction of the reference block. Obtain the first picture block, and from the reference frame in the second direction of the reference block, obtain the second picture block indicated by the initial motion vector in the second direction of the reference block, and obtain the temporal of the reference block. It is configured to perform a weighting process on the first picture block and the second picture block in order to acquire the prediction block.

In a feasible implementation, acquisition module 2302 is specifically configured to use the predicted motion vector of the picture block to be processed as the initial motion vector of the picture block to be processed.

In a feasible implementation, acquisition module 2302 adds the predicted motion vector of the picture block to be processed and the motion vector difference of the picture block to be processed to the initial of the picture block to be processed. It is specifically configured to get a motion vector.

In a feasible implementation, the device 2300 is used for video decoding and the motion vector differences of the picture blocks to be processed are obtained by analyzing the first discriminant information in the bitstream.

In a feasible implementation, device 2300 is used for video decoding and decision module 2301 analyzes the bitstream to obtain second identification information and is based on the second identification information of the picture block to be processed. It is specifically configured to determine the reference block.

In a feasible implementation, device 2300 is used for video coding and the decision module 2301 is taken from one or more candidate reference blocks of the picture block to be processed as a reference block of the picture block to be processed. It is specifically configured to select the candidate reference block with the minimum rate distortion cost.

FIG. 24 is a schematic block diagram of a video coding device according to an embodiment of the present application. The device 2400 may be applied to the encoder or may be applied to the decoder. Device 2400 includes processor 2401 and memory 2402. The processor 2401 and the memory 2402 are connected to each other (eg, via bus 2404). In a possible implementation, the device 2400 may further include a transceiver 2403. Transceiver 2403 is connected to processor 2401 and memory 2402 and is configured to receive / transmit data.

The memory 2402 is a random access memory (RAM), a read-only memory (ROM), and an erasable programmable. Includes, but is not limited to, read-only memory (EPROM) or compact disc read-only memory (CD-ROM). Memory 2402 is configured to store relevant program code and video data.

Processor 2401 may be one or more central processing units (CPUs). When the processor 2401 is one CPU, the CPU may be a single core CPU or a multi-core CPU.

Processor 2401 is configured to read the program code stored in memory 2402 and perform any implementation solution corresponding to FIGS. 13 to 20 and operations in various feasible implementations of the implementation solution. To.

For example, embodiments of the present application further provide a computer-readable storage medium. A computer-readable storage medium stores instructions. When the instruction is executed on the computer, the computer
It makes it possible to perform operations in any implementation solution corresponding to FIGS. 13 to 20 and in various implementable implementations of the implementation solution.

For example, embodiments of the present application further provide computer program products that include instructions. When a computer program product is run on a computer, the computer is capable of performing any implementation solution corresponding to FIGS. 13-20, as well as operations in various implementable implementations of the implementation solution. To.

FIG. 25 shows a motion vector acquisition device according to an embodiment of the present application. The apparatus includes a determination module 2501, a first acquisition module 2502, and a second acquisition module 2503.

The decision module 2501 is configured to perform step 1904.

The first acquisition module 2502 is configured to perform step 1907.

The second acquisition module 2503 is configured to perform step 1907.

Optionally, the reference block is in the preset range so that the coding tree block CTB where the reference block is located and the coding tree block where the picture block to be processed is located are on the same line. Including being located.
Correspondingly, the fact that the reference block is out of the preset range is related to step 1907, that is, the coding tree block in which the reference block is located and the coding tree in which the picture block to be processed is located. -Includes that the block is located on a different line.

Optionally, having the reference block in the preset range includes that the reference block and the picture block to be processed are in the same coding tree block.
Correspondingly, the fact that the reference block is out of the preset range includes that the reference block and the picture block to be processed are located in different coding tree blocks in relation to step 1907.

As an option
The fact that the reference block is within the preset range means that the coding tree block in which the reference block is located is the same as the coding tree block in which the picture block to be processed is located, or the reference block is located. Includes that the located coding tree block is a left-adjacent or right-adjacent block of the coding tree block in which the picture block to be processed is located.
Correspondingly, the fact that the reference block is out of the preset range means that for step 1907, the coding tree block in which the reference block is located is the same as the coding tree block in which the picture block to be processed is located. Includes that the coding tree block in which the reference block is located is not the left-adjacent or right-adjacent block of the coding tree block in which the picture block to be processed is located.

FIG. 26 shows a motion vector residual determination device according to an embodiment of the present application. The apparatus includes an analysis module 2601, an addition module 2602, a determination module 2603, and an acquisition module 2604.

The analysis module 2601 is configured to perform step 2001.

Addition module 2602 is configured to perform step 2002.

The decision module 2603 is configured to perform step 1422.

Acquisition module 2604 is configured to perform step 2008.

Optionally, the device is used for bidirectional inter-prediction, the final motion vector contains a first final motion vector and a second final motion vector, and the initial motion vector is the first initial motion vector. Includes motion vector and second initial motion vector; first final motion vector and first initial motion vector are motion compensation blocks based on the first reference frame list of the picture block to be processed. The second final motion vector and the second initial motion vector indicate the motion compensation block based on the second reference frame list of the picture block to be processed; and the final motion vector and Using the difference from the initial motion vector as the motion vector residuals of the picture block being processed is:
The difference between the first final motion vector and the first initial motion vector is used as the first motion vector residual of the picture block to be processed.

FIG. 27 shows a motion vector data storage device according to an embodiment of the present application. The apparatus includes a first analysis module 2701, an acquisition module 2702, a second analysis module 2703, and a storage module 2704.

The first analysis module 2701 is configured to perform step 2001.

Acquisition module 2702 is configured to perform step 2007.

The second analysis module 2703 is configured to perform step 2001.

Storage module 2704 is configured to perform step 2008.

Those skilled in the art will appreciate that in combination with the examples described in the embodiments disclosed herein, the units and algorithm steps are implemented by electronic hardware, or a combination of computer software and electronic hardware. Will recognize that is possible. Whether a function is performed by hardware or software depends on the design constraints of the particular application and technical solution. One of ordinary skill in the art can use various methods to implement the described functionality for each particular application, but such implementation should be considered to be beyond the scope of this application. No.

It will be apparent to those skilled in the art that for convenience and brief description purposes, those skilled in the art should refer to the corresponding processes in embodiments of the methods described above for detailed operating processes of the systems, devices and units described above. Will be done. Details will not be explained here again.

All or part of the aforementioned embodiments can be realized by software, hardware, firmware, or any combination thereof. When the software is used to implement an embodiment, the embodiment can be fully or partially implemented in the form of a computer program product. Computer program products include one or more computer instructions. When computer program instructions are loaded and executed on a computer, all or part of the procedure or function arises according to embodiments of the present application. The computer may be a general purpose computer, a dedicated computer, a computer network, or other programmable device. Computer instructions may be stored on a computer-readable storage medium or may be transmitted from a computer-readable storage medium to another computer-readable storage medium. For example, a computer command can be wired (eg, coaxial cable, fiber optic, or digital subscriber line) or from a website, computer, server, or data center to another website, computer, server, or data center. It can be transmitted wirelessly (eg, infrared, microwave, etc.). The computer-readable storage medium may be any usable medium accessible by the computer, or a data storage device such as a server or data center that integrates one or more usable media. The medium that can be used may be a magnetic medium (for example, a floppy disk, a hard disk, or a magnetic tape), an optical medium (for example, a digital versatile disk), a semiconductor medium (for example, a solid-state drive), or the like.

In the aforementioned embodiments, the description in each embodiment has an individual focus. For parts not described in detail in the embodiment, refer to the related description in the other embodiments.

The above description merely describes a particular implementation of the present application and is not intended to limit the scope of protection of the present application. Any modifications or substitutions readily apparent to those skilled in the art within the technical scope disclosed herein shall be within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be in accordance with the scope of protection of the claim.

This application claims priority to Chinese Patent Application No. 201811020181.9, entitled "Video Coding Methods and Devices" filed September 3, 2018, which is incorporated herein by reference in its entirety.

This application claims priority to Chinese Patent Application No. 201811271726.3 entitled "Motion Vector Acquisition Methods and Devices, Computer Devices, and Storage Media" filed October 29, 2018, in its entirety. Is incorporated herein by reference.

Technical Fields The present application relates to the field of video compression technology, in particular motion vector acquisition methods and devices, computer devices, and storage media.

Background Video-related applications are becoming more and more popular in everyday life. With the maturity of computer technology, video processing technology has also evolved significantly. Video coding technology has evolved tremendously. Redundancy can be eliminated as much as possible by utilizing the correlation between the coding unit and the prediction unit when the coding unit within the frame of the picture is encoded. At the time of decoding, the information acquired after the redundancy is removed is decoded to acquire the picture information corresponding to the coding unit.

The decoding process may be as follows: After acquiring the motion information of the coding unit, the decoder sets a list of candidate predicted motion vectors based on the motion information and the acquired motion. Coding by selecting the optimal predicted motion vector from the list of candidate predicted motion vectors based on the information and using the reference unit indicated by the optimal predicted motion vector as the starting point. Search for the prediction unit that most closely resembles the unit and update the best predicted motion vector in the list of candidate predicted motion vectors to the motion vector of the prediction unit (ie, the updated predicted motion). The vector is the predicted motion vector of the coding unit), and the picture information corresponding to the coding unit is acquired based on the updated predicted motion vector and the prediction unit.

In the process of realizing the present application, the inventor has found that the related technique has at least the following drawbacks : the decoder may decode multiple coding units simultaneously during decoding. However, if the predicted motion vector of the current coding unit needs to be used by another coding unit, decoding will update the optimal predicted motion vector of the current coding unit. It can only be executed after it has been done, and as a result, delays are very likely to occur.

Embodiments of the present application provide motion vector acquisition methods and devices, computer devices, and storage media to solve the decoding delay problem in related arts.

According to one aspect, the present application provides a motion vector acquisition method, which is:
A step that determines the reference block of the picture block to be processed, in which the reference block and the picture block to be processed are located in the same frame of the picture.
If the reference block is within a preset range, it is a step to get the motion vector of the picture block to be processed based on the initial motion vector of the reference block, and the preset range is processed. Steps and steps that are determined based on the position of the target picture block,
This is the step to get the motion vector of the picture block to be processed based on the final motion vector of the reference block when the reference block is out of the preset range, and the final motion vector is the initial. Includes steps, which are obtained based on the motion vector of.

In a possible implementation, the fact that the reference block is within the preset range means that the coding tree block CTB where the reference block is located and the coding tree block where the picture block to be processed is located are on the same line. Including being located in
Correspondingly, the fact that the reference block is out of the preset range means that the coding tree block where the reference block is located and the coding tree block where the picture block to be processed is located are located on different lines. Including doing.

Based on the possible implementations described above, different preset ranges may be provided so that the motion vector of the picture block to be processed is determined based on the different preset ranges. It is possible to select multiple motion vectors of the picture block to be processed.

In a possible implementation, the coding tree block where the reference block is located and the coding tree block where the picture block to be processed is located are on different lines, and the coding tree block where the reference block is located is , Above or in the upper left of the coding tree block where the picture block to be processed is located.

Based on the possible implementations described above, it is possible to provide another pre-set range so that multiple pre-set conditions can be provided for motion vector selection.

In a possible implementation, the fact that the reference block is within the preset range includes that the reference block and the picture block to be processed are located in the same coding tree block.
Correspondingly, the fact that the reference block is out of the preset range includes that the reference block and the picture block to be processed are located in different coding tree blocks.

Multiple possible specific pre-set ranges may be provided based on the possible implementations described above.

In a possible implementation, the fact that the reference block is within the preset range means that the coding tree block in which the reference block is located is the same as the coding tree block in which the picture block to be processed is located. , Or the coding tree block in which the reference block is located is a left-adjacent or right-adjacent block of the coding tree block in which the picture block to be processed is located, and correspondingly, the reference block is pre-located. Being out of the set range means that the coding tree block in which the reference block is located is not the same as the coding tree block in which the picture block to be processed is located, or the coding tree block in which the reference block is located. Includes that the tree block is not the left or right adjacency block of the coding tree block in which the picture block to be processed is located.

Based on the possible implementations described above, a plurality of possible specific pre-set ranges may be provided.

In a possible implementation, the step of determining the reference block of the picture block to be processed is:
A step of continuously determining one or more pre-set candidate reference blocks as reference blocks in a pre-set order, where the candidate reference block is a pre-set space with the picture block to be processed. Includes steps, including picture blocks that have a relative positional relationship.

Based on the above implementation, the reference block of the picture block to be processed may be determined, and then the initial motion vector of the picture block to be processed may be determined based on the reference block.

In a possible implementation, the step of determining the reference block of the picture block to be processed is:
The step of parsing the bitstream to obtain one or more first identification information,
A step of determining a reference block from a plurality of candidate reference blocks of a picture block to be processed based on one or more first identification information, wherein the candidate reference block is in advance with the picture block to be processed. Includes steps, including picture blocks with set spatial positional relationships.

Based on the possible implementation described above, the reference block of the picture block to be processed can be determined based on the identification information in the bitstream, and then the initial motion vector of the picture block to be processed. Can be determined based on the reference block.

In a possible implementation, it is possible that the final motion vector is obtained based on the initial motion vector.
To get multiple candidates for the final motion vector by adding the initial motion vector and multiple preset offset vectors separately,
This includes determining the final motion vector candidate corresponding to the minimum distortion cost among the plurality of final motion vector candidates as the final motion vector.

Based on the possible implementations described above, the final motion vector candidate corresponding to the lowest strain cost is used as the final motion vector, resulting in a more accurate final motion vector.

In a possible implementation, the method is used for bidirectional inter-prediction, the final motion vector contains a first final motion vector and a second final motion vector, and the initial motion vector is the first. The initial motion vector and the second initial motion vector are included, and the first final motion vector and the first initial motion vector are based on the first reference frame list of the picture block to be processed. The second final motion vector and the second initial motion vector indicate the motion compensation block based on the second reference frame list of the picture block to be processed.
That the final motion vector is obtained based on the initial motion vector,
To obtain multiple candidates for the first final motion vector by adding the first initial motion vector and multiple preset offset vectors separately.
The candidate for the first final motion vector corresponding to the minimum strain cost among the plurality of candidates for the first final motion vector is determined as the first final motion vector. The first final motion vector corresponds to the first offset vector in a plurality of preset offset vectors.
To obtain the second offset vector, the size of the second offset vector is equal to that of the first offset vector, and the direction of the second offset vector is opposite to that of the first offset vector. There is, that,
It involves adding a second initial motion vector and a second offset vector to obtain a second final motion vector.

Based on the possible implementations described above, the final motion vector is obtained based on the initial motion vector in bidirectional prediction mode.

In a possible implementation, the method is used for bidirectional inter-prediction, the final motion vector contains a first final motion vector and a second final motion vector, and the initial motion vector is the first. The initial motion vector and the second initial motion vector are included, and the first final motion vector and the first initial motion vector are based on the first reference frame list of the picture block to be processed. The second final motion vector and the second initial motion vector indicate the motion compensation block based on the second reference frame list of the picture block to be processed.
That the final motion vector is obtained based on the initial motion vector,
To obtain multiple candidates for the first final motion vector by adding the first initial motion vector and multiple preset offset vectors separately.
Determining the candidate for the first final motion vector corresponding to the minimum strain cost among the plurality of candidates for the first final motion vector as the first final motion vector,
To obtain multiple candidates for the second final motion vector by adding the second initial motion vector and the plurality of pre-set second offset vectors separately.
This includes determining the candidate for the second final motion vector corresponding to the minimum strain cost among the plurality of candidates for the second final motion vector as the second final motion vector.

Based on the above possible implementation, another method of obtaining the final motion vector based on the initial motion vector in the bidirectional prediction mode is provided, resulting in the final motion vector in the bidirectional prediction mode. It is possible to provide multiple ways to obtain.

According to the second aspect, the present application provides a method for determining a motion vector residual, which is:
The step of analyzing the bitstream to obtain the second identification information, which is used to determine the initial motion vector of the picture block to be processed.
The step of adding the initial motion vector and multiple pre-set offset vectors separately to get multiple candidates for the final motion vector,
A step of determining the final motion vector candidate corresponding to the minimum distortion cost among multiple candidates of the final motion vector as the final motion vector, and
Either the difference between the final motion vector and the initial motion vector is used as the motion vector residual of the picture block to be processed, or the final motion vector is the motion vector of the picture block to be processed. Includes steps to use as residuals.

In a possible implementation, the method is used for bidirectional inter-prediction, the final motion vector contains a first final motion vector and a second final motion vector, and the initial motion vector is the first. The initial motion vector and the second initial motion vector are included, and the first final motion vector and the first initial motion vector are based on the first reference frame list of the picture block to be processed. The second final motion vector and the second initial motion vector indicate the motion compensation block based on the second reference frame list of the picture block to be processed.
The step of using the difference between the final motion vector and the initial motion vector as the motion vector residuals of the picture block to be processed is:
It comprises the step of using the difference between the first final motion vector and the first initial motion vector as the first motion vector residual of the picture block to be processed.

Based on the possible implementation described above, the motion vector residuals of the picture block to be processed can be determined in the bidirectional interprediction mode.

In a possible implementation, the step of using the difference between the final motion vector and the initial motion vector as the motion vector residuals of the picture block being processed is:
Further included is the step of using the difference between the second final motion vector and the second initial motion vector as the second motion vector residual of the picture block to be processed.

In a possible implementation, the step of using the final motion vector as the motion vector residuals of the picture block to be processed is:
It comprises the step of using the first final motion vector as the first motion vector residual of the picture block to be processed.

In a possible implementation, the step of using the final motion vector as the motion vector residuals of the picture block to be processed is:
It further comprises the step of using the second final motion vector as the second motion vector residual of the picture block to be processed.

Based on the multiple possible implementations described above, multiple methods are provided to determine the motion vector residuals of the picture block to be processed so that the encoder and decoder have coding efficiency based on the motion vector residuals. And the decoding efficiency can be improved.

According to a third aspect, the present application provides a motion vector data storage method, which is:
The step of analyzing the bitstream to obtain the second and third identification information, which is used to determine the initial predicted motion vector of the picture block to be processed. To be done, steps and
Steps to get the final predicted motion vector based on the initial predicted motion vector and multiple pre-set offset vectors,
When the third identification information indicates that the bitstream carries the motion vector residuals of the picture block to be processed, the bitstream is analyzed to obtain the motion vector residuals, and the motion vector residuals are obtained. Steps to store in the target storage space and
If the third identification information indicates that the bitstream does not carry the motion vector residuals of the picture block being processed, then between the final predicted motion vector and the initial predicted motion vector. Includes the steps of storing the differences in the target storage space or storing the final predicted motion vector in the target storage space.

In a possible implementation, the third identification information indicates that the bitstream carries the motion vector residuals of the picture block to be processed.
The third identification information includes indicating that the prediction mode of the picture block to be processed is the AMVP mode.

Based on the possible implementation described above, the motion vector data of the picture block to be processed can be stored in multiple prediction modes.

In a possible implementation, the third identification information indicates that the bitstream does not carry the motion vector residuals of the picture block being processed.
The third identification information includes indicating that the prediction mode of the picture block to be processed is the merge mode or the skip mode.

Based on the possible implementations described above, different identification information is used to distinguish between different prediction modes, so that the motion vector data of the picture block being processed is stored in multiple prediction modes. Is possible.

In a possible implementation, the step of getting the final predicted motion vector based on the initial predicted motion vector and multiple pre-set offset vectors is
The step of adding the initial motion vector and multiple pre-set offset vectors separately to get multiple candidates for the final motion vector,
It includes a step of determining a final motion vector candidate corresponding to the minimum distortion cost among a plurality of final motion vector candidates as a final motion vector.

Based on the above possible implementation, the distortion cost is used to obtain the final predicted motion vector with higher accuracy, and the stored motion vector data is more accurate.

According to the fourth aspect, the present application provides a motion vector acquisition device configured to execute the above motion vector acquisition method. Specifically, the motion vector acquisition device includes a functional module configured to perform the motion vector acquisition method provided in any one of the first aspect or the optional method of the first aspect. .. The above aspect corresponds to the motion vector acquisition method.

According to a fifth aspect, the present application provides a motion vector residual determination device configured to perform the motion vector residual determination method described above. Specifically, the motion vector residual determination device is configured to perform the motion vector residual determination method provided by any one of the second aspect or the optional method of the second aspect. Includes functional modules. The above aspect corresponds to the motion vector residual determination method.

According to a sixth aspect, the present application provides a motion vector data storage device configured to perform the motion vector data storage method described above. Specifically, the motion vector data storage device is configured to perform the motion vector data storage method provided by any one of the third aspect or the optional method of the third aspect. Includes functional modules that have been created. The above aspect corresponds to the motion vector data storage method.

According to a seventh aspect, the present application provides a computer device. Computer devices include processors and memory. The memory stores at least one instruction, which is loaded and executed by the processor to implement the operations performed in the motion vector acquisition method described above.

According to an eighth aspect, the present application provides a computer device. Computer devices include processors and memory. The memory stores at least one instruction, which is loaded and executed by the processor to perform the operations performed in the motion vector data storage method described above.

According to a ninth aspect, the present application provides a computer device. Computer devices include processors and memory. The memory stores at least one instruction, which is loaded and executed by the processor to implement the operations performed in the motion vector residual determination method described above.

According to a tenth aspect, the present application provides a computer-readable storage medium. The storage medium stores at least one instruction, which is loaded and executed by the processor to realize the operations performed in the motion vector acquisition method described above.

According to the eleventh aspect, the present application provides a computer-readable storage medium. The storage medium stores at least one instruction, which is loaded and executed by the processor to implement the operations performed in the motion vector residual determination method described above.

According to a twelfth aspect, the present application provides a computer-readable storage medium. The storage medium stores at least one instruction, which is loaded and executed by the processor to realize the operations performed in the motion vector data storage method described above.

The technical solutions provided in this application include at least the following beneficial effects:

The initial motion vector of the picture block to be processed is determined by using the positional relationship between the reference block and the picture block to be processed. If the reference block and the picture block to be processed are within a preset range, the decoder uses the initial motion vector of the reference block as the motion vector of the picture block to be processed. If the reference block and the picture block to be processed are out of the preset range, the decoder uses the final motion vector of the reference block as the motion vector of the picture block to be processed. In this case, if the picture block to be processed needs to be decoded by using the final motion vector of the reference block at the time of decoding, the decoder will use the picture block to be processed. The initial motion vector of the reference block can be used as the motion vector of the picture block to be processed so that the block can be used, which gives the final motion vector of the reference block. It is possible to avoid the case where the motion vector of the picture block to be processed can be acquired and improve the decoding efficiency only after the motion vector is obtained.

FIG. 3 is a block diagram of an example video coding system that can be configured for use in embodiments of the present application.

FIG. 3 is a system block diagram of an example video encoder that can be configured for use in embodiments of the present application.

FIG. 3 is a system block diagram of an example video decoder that can be configured for use in embodiments of the present application.

It is a schematic block diagram of an example of an inter-prediction module that can be configured for use in embodiments of the present application.

It is a schematic diagram of an example of a coding unit and adjacent picture blocks associated with the coding unit.

It is a flowchart of an implementation example which constructs a list of predicted motion vectors of candidates.

It is a schematic of the implementation example which adds the combined candidate motion vector to the list of predicted motion vectors of merge mode candidates.

It is a schematic of the implementation example which adds a scaled candidate motion vector to a list of predicted motion vectors of merge mode candidates.

It is a schematic of the implementation example which adds a zero motion vector to a list of predicted motion vectors of merge mode candidates.

It is a flowchart of the implementation example of the merge prediction mode.

It is a flowchart of the implementation example of the advanced motion vector prediction mode.

FIG. 3 is a flow chart of an example motion compensation implementation performed by a video decoder that can be configured for use in embodiments of the present application.

It is a schematic flowchart of the motion vector update method during video coding by embodiment of this application.

It is a schematic flowchart of the motion vector update method during video decoding by embodiment of this application.

It is a schematic flowchart which updates a motion vector by embodiment of this application.

It is a schematic flowchart which updates a motion vector by embodiment of this application.

It is a schematic flowchart which updates a motion vector by embodiment of this application. It is a schematic flowchart which updates a motion vector by embodiment of this application.

It is a schematic flowchart which updates a motion vector by embodiment of this application.

It is a schematic flowchart of the motion vector acquisition method by embodiment of this application.

It is a schematic flowchart of the motion vector acquisition method by embodiment of this application.

It is a schematic diagram which selects the motion vector within the range specified according to the embodiment of this application.

It is a schematic diagram of the acquisition method of the present prediction block by embodiment of this application.

It is a schematic block diagram of the motion vector acquisition apparatus during video decoding by the embodiment of this application.

FIG. 3 is a schematic block diagram of a video coding device according to an embodiment of the present application.

It is a structural drawing of the motion vector acquisition apparatus by embodiment of this application.

It is a structural drawing of the motion vector residual determination apparatus by embodiment of this application.

It is a structural diagram of the motion vector data storage apparatus according to the embodiment of this application.

In order to further clarify the objectives, technical solutions, and advantages of the present application, the implementation of the present application will be described in more detail below with reference to the accompanying drawings.

FIG. 1 is a schematic block diagram of a video coding system 10 according to an embodiment of the present application. As shown in FIG. 1, the system 10 includes a source device 12. The source device 12 produces encoded video data that should later be decoded by the destination device 14. The source device 12 and the destination device 14 are desktop computers, notebook computers, tablet computers, set-top boxes, telephone handset such as "smart" phones, "smart" touchpads, televisions, cameras, display devices. , Digital media players, video game consoles, video streaming transmitters, etc. may include any of a wide range of devices. In some applications, the source device 12 and the destination device 14 may be equipped for wireless communication.

The destination device 14 can receive the encoded video data to be decoded via the link 16. Link 16 may include any type of medium or device capable of transmitting encoded video data from the source device 12 to the destination device 14. In a feasible implementation, the link 16 may include a communication medium that allows the source device 12 to send the encoded video data directly to the destination device 14 in real time. The encoded video data can be modulated according to a communication standard (eg, a wireless communication protocol) and then transmitted to the destination device 14. The communication medium may include any wireless or wired communication medium, such as a radio frequency spectrum or at least one physical transmission line. The communication medium can form part of a packet-based network (eg, a local area network, a wide area network, or a global network such as the Internet). The communication medium may include routers, switches, base stations, or any other device that can be used to facilitate communication from the source device 12 to the destination device 14.

Alternatively, the encoded data may be output from the output interface 22 to the storage device. Similarly, the encoded data may be accessed from the storage device via the input interface. The storage device is a plurality of distributed data storage media or locally accessible data storage media such as hard drives, Blu- rays, digital versatile discs (DVDs), compact discs read-only. Memory (compact disc read-only memory, CD-ROM), flash memory, volatile or non-volatile memory, or any other suitable digital storage configured to store encoded video data. It may contain any one of the media. In another feasible implementation, the storage device may correspond to a file server or another intermediate storage device capable of holding the encoded video produced by the source device 12. The destination device 14 can access the stored video data from the storage device by streaming transmission or download. The file server can be any type of server capable of storing the encoded video data and transmitting the encoded video data to the destination device 14. In a feasible implementation, file servers include website servers, file transfer protocol servers, network-attached storage devices, or local disk drives. The destination device 14 can access the encoded video data via any standard data connection, including an internet connection. A data connection is a wireless channel (eg, a wireless-fidelity (Wi-Fi) connection) or a wired connection (eg, a wireless-fidelity) suitable for accessing encoded video data stored on a file server. Cable modems), or combinations thereof. The transmission of the encoded video data from the storage device may be a streaming transmission, a download transmission, or a combination thereof.

The techniques in this application are not necessarily limited to wireless applications and settings. The technology is used in multiple multimedia applications such as over-the-air television broadcasting, cable television transmission, satellite television transmission, video streaming transmission (eg, over the Internet), and data storage media. Applicable to video decoding to support any of digital video encoding for storage, decoding of digital video stored on data storage media, or any other application. Is possible. In some feasible implementations, the system 10 will support one-way or two-way video transmission to support applications such as video streaming transmission, video playback, video broadcasting, and / or video telephone. It may be configured in.

In the feasible implementation of FIG. 1, the source device 12 includes a video source 18, a video encoder 20, and an output interface 22. In some applications, the output interface 22 may include a modulator / demodulator (modem) and / or a transmitter. In the source device 12, the video source 18 may include, for example, the following sources: a video capture device (eg, a video camera), a video archive containing previously captured video, and a video receiving video from a video content provider. It can include a feed-in interface and / or a computer graphics system that produces computer graphics data as source video, or a combination thereof. In a feasible implementation, if the video source 18 is a video camera, the source device 12 and the destination device 14 can configure a camera phone or a video phone. For example, the techniques described herein can be applied, for example, to video decoding and can be applied to wireless and / or wired applications.

The video encoder 20 can encode captured or pre-captured video, or computer-generated video. The encoded video data can be transmitted directly to the destination device 14 via the output interface 22 of the source device 12. The encoded video data can also (or alternatively) be stored in the storage device 24 so that the destination device 14 or another device can subsequently be decoded and / or played back. To access the encoded video data.

The destination device 14 includes an input interface 28, a video decoder 30, and a display device 32. In some applications, the input interface 28 may include a receiver and / or a modem. The input interface 28 of the destination device 14 receives the encoded video data via the link 16. The encoded video data transmitted or provided to the storage device 24 over the link 16 is generated by the video encoder 20 and for decoding the video data by a video decoder such as the video decoder 30. May contain multiple syntax elements used in. These syntax elements can be included in the encoded video data transmitted over the communication medium and stored on the storage medium or file server.

The display device 32 may be integrated with the destination device 14 or may be arranged outside the destination device 14. In some feasible implementations, the destination device 14 may include an integrated display device or may be configured to connect to an interface of an external display device. In other feasible implementations, the destination device 14 may be a display device. Generally, the display device 32 displays the decoded video data to the user and is among a plurality of display devices such as a liquid crystal display, a plasma display, an organic light emitting diode display, or another type of display device. Any one can be included.

The video encoder 20 and the video decoder 30 can operate according to, for example, the next generation video coding compression standard (H.266) currently being developed, and H.I. It may be compliant with the 266 test model.

Alternatively, the video encoder 20 and the video decoder 30 are described, for example, by ITU-T H. 265 standard or ITU-T H. It may operate according to other specialized or industrial standards such as the 264 standard, or extensions of these standards. ITU-T H. The 265 standard, also referred to as the high efficiency video decoding standard, is the ITU-T H.D. Alternatively, the 264 standard is also referred to as moving picture experts group (MPEG-4) Part 10 or advanced video coding (AVC). However, the technique of the present application is not limited to any particular decoding standard. In other feasible implementations, video compression standards are MPEG-2 and ITU-TH. 263 is included.

Although not shown in FIG. 1, in some embodiments, the video encoder 20 and the video decoder 30 may be integrated with the audio encoder and audio decoder, respectively, and may be the same data stream or separate. Appropriate multiplexer / demultiplexer (multiplexer-demultiplexer, MUX-DEMUX) units or other hardware and software may be included to encode both audio and video in the data stream. If applicable, in some feasible implementations, the MUX-DEMUX unit is an ITU H. It may comply with the 223 multiplexer protocol, or other protocols such as the user datagram protocol (UDP).

Video encoder 20 and video decoder 30, a plurality of suitable encoder circuitry, for example, one or more microprocessors, digital signal processor (digital signal processor, DSP), application specific integrated circuits (application specific integrated circuit, ASIC ), Field programmable gate array (FPGA), discrete logic, software, hardware, firmware, or any combination thereof. be. When some techniques are realized as software, the device stores software instructions in a suitable non-temporary computer-readable medium and uses one or more processors to realize the techniques in this application. By doing so, the instruction can be executed in the form of hardware. The video encoder 20 and the video decoder 30 may be included in one or more encoders or decoders, respectively. Both the video encoder 20 and the video decoder 30 can be integrated as part of a combined encoder / decoder (CODEC) in the corresponding device.

The present application may relate, for example, to another device in which the video encoder 20 "signals" certain information to, for example, the video decoder 30. However, it should be understood that the video encoder 20 can associate a particular syntax element with an encoded portion of the video data in order to signal the information. That is, the video encoder 20 can store a specific syntax element in the header information of the encoded portion of the video data in order to signal the data. In some applications, these syntax elements are encoded and stored (eg, stored in a storage system or file server) before being received and decoded by the video decoder 30. Thus, the term "signal" is used, for example, to transmit or compress a syntax, regardless of whether the transmission takes place in real time, near real time, or within a time span. It may mean the transmission of other data used to decode the video data. For example, transmission can be performed if the syntax element is stored on the medium during coding, after which the syntax element is decoded at any time after it is stored on the medium. It can be taken out by a computer.

The joint video team (JVT) is H.K. 265 High efficiency video coding (HEVC) standard has been developed. The HEVC standardization is based on an evolved model of the video decoder, the model being called the HEVC test model (HM) . The latest H. The 265 standard document is http: // www. itu. int / rec / T-REC-H. Available at 265. The latest version of the standard document is H. 265 (12/16), the standard document is incorporated herein by reference in its entirety. In HM, the video decoder is ITU-TH. It is hypothesized to have some additional capabilities associated with existing 264 / AVC algorithms. For example, H. The 264 can provide 9 intra-predictive coding modes, while the HM can provide up to 35 intra-predictive coding modes.

JVET is H. We are working on the development of the 266 standard. H. The 266 standardization process is based on an evolved model of the video decoder, which is based on H.D. It is called the 266 test model. H. The explanation of the 266 algorithm is http: // phenix. int−evry. It is available on fr / jvet and the latest algorithm description is contained in JVET-G1001-v1. The entire algorithm description document is incorporated herein by reference in its entirety. In addition, reference software for the JEM test model is available at https: // jvet. hhi. fraunhofer. It is available at de / svn / svn_HMJEM Software / and is incorporated by reference in its entirety into this application.

Generally, as described in the HM working model, a video frame or picture is divided into a series of tree blocks or a largest coding unit (LCU) containing both Luma and Chroma samples. It is possible to be done. The LCU is also called a coding tree unit (CTU). The tree block is H. It has the same function as that of the 264 standard macroblock. The slice contains several contiguous tree blocks in the order of decryption. The video frame or picture can be divided into one or more slices. Each tree block can be divided into coding units based on quadtrees. For example, a tree block that acts as the root node of a quadtree can be split into four child nodes, each child node can also act as a parent node, and the other four. Divided into two child nodes. The final indivisible child node that acts as the leaf node of the quadtree contains a decryption node, eg, a decrypted video block. In the syntax data associated with the decrypted bitstream, the maximum number of times the tree block can be split and the minimum size of the decrypted node may be determined.

It should be noted that one CTU contains one coding tree block (CTB) and two chroma CTBs at the same location and several corresponding syntax elements. The CTB can be used directly as a coding block (CB), or multiple small coding blocks CB, one Luma CB, two Chroma CBs, and several. It is possible to divide into the corresponding syntax elements of the above in the form of a quadtree to form one coding unit (CU).

The CU includes a decoding node, a prediction unit (PU), and a transform unit (TU) associated with the decoding node. The size of the CU corresponds to the size of the decoding node, and the shape of the CU needs to be square. The size of the CU may range from 8x8 pixels up to 64x64 pixels, or may be a larger tree block size. Each CU can include one or more PUs and one or more TUs. For example, CU-related syntax data can describe dividing one CU into one or more PUs. The partitioning pattern can be different if the CU is coded in skip or direct mode, in intra-prediction mode, or in inter-prediction mode. The PU obtained by partitioning may have a non-square shape. For example, CU-related syntax data can also describe the partitioning of a CU to one or more TUs based on a quadtree. The TU may have a square or non-square shape.

The HEVC standard allows TU-based conversion. Different CUs may contain different TUs. The size of the TU is usually set based on the size of the PU in a given CU defined for the partitioned LCU. However, this may not always be the case. The size of the TU is usually the same as or smaller than that of the PU. In some feasible implementations, a quadtree structure called a "residual quadtree" (RQT) is used to divide the CU-corresponding residual sample into smaller units. there is a possibility. Leaf nodes of RQT are sometimes referred to as TUs. Pixel differences associated with the TU may be converted to generate a conversion factor, and the conversion factor may be quantized.

Generally, the PU contains data related to the prediction process. For example, if the PU is encoded in intra mode, the PU may contain data describing the intra prediction mode of the PU. In another feasible implementation, if the PU is encoded in intermode, the PU may contain data defining motion vectors for the PU. For example, the data defining the motion vector for the PU is indicated by the horizontal component of the motion vector, the vertical component of the motion vector, the resolution of the motion vector (eg, 1/4 pixel accuracy or 1/8 pixel accuracy), and the motion vector. A reference picture and / or a motion vector reference picture list (eg, list 0, list 1, or list C) can be described.

Generally, transformation and quantization processes are used for TUs. A given CU containing one or more PUs can also contain one or more TUs. After performing the prediction, the video encoder 20 can calculate the residuals corresponding to the PU. The residuals include pixel differences. Pixel differences can be converted to conversion factors, which are quantized and scanned using the TU to produce a serialized conversion factor for entropy decoding. In the present application, the term "video block" is typically used to indicate a CU decoding node. In some specific applications, the term "video block" can also be used to indicate a tree block containing a decryption node, PU, and TU, such as an LCU or CU.

A video sequence usually includes a series of video frames or pictures. For example, group of pictures (group of picture s, GOP) includes a series of video pictures, or one or more video pictures. The GOP may include syntax data in the header information of the GOP, the header information of one or more pictures, and the like, and the syntax data describes the amount of pictures contained in the GOP. Each slice of a picture may contain slice syntax data that describes the coding mode of the corresponding picture. The video encoder 20 typically performs an operation on a video block within an individual video slice to encode the video data. The video block may correspond to a decryption node in the CU. The size of the video block may be fixed or variable and may vary depending on the specified decoding standard.

In a feasible implementation, HM supports prediction at various PU sizes. Assuming that the size of a particular CU is 2N × 2N, the HM will make an intra prediction for the PU using a size of 2N × 2N or N × N and 2N × 2N, 2N × N, N × 2N, or N. Supporting inter-prediction for symmetric PUs using xN size, HM also supports inter-prediction asymmetric parties for PUs using xN x nU, 2N x nD, nL x 2N, or nR x 2N sizes. It also supports shining. In asymmetric partitioning, the CU is not divided in one direction, but in two directions, one part occupying 25% of the CU and the other part occupying 75% of the CU. The portion that occupies 25% of the CU is indicated by an indicator that indicates "n" followed by "U (Up)", "D (Down)", "L (Left)", and "R (Right)". Therefore, for example, "2N x nU" refers to a horizontally divided 2N x 2N CU, the upper part is 2N x 0.5N PU, and the lower part is 2N x 1.5N PU.

In the present application, "NxN" and "N times N" indicate the pixel size of a vertical and horizontal dimensional video block, such as 16x16 pixels or the number of pixels obtained by multiplying 16 by 16. It can be used interchangeably for. Generally, a 16x16 block has 16 pixels (y = 16) in the vertical direction and 16 pixels (x = 16) in the horizontal direction. Similarly, an N × N block usually has N pixels in the vertical direction and N pixels in the horizontal direction, where N is a non-negative integer value. The pixels in the block can be arranged in rows and columns. Also, the amount of pixels in the horizontal direction and the amount of pixels in the vertical direction of the block may not always be the same. For example, the block may contain N × M pixels, where M is not necessarily equal to N.

After performing intra-or inter-predictive decoding for the PU in the CU, the video encoder 20 can calculate the residual data of the TU in the CU. The PU may contain pixel data within the spatial domain (also known as the pixel domain), and the TU is a transformation applied to the residual video data (eg, the discrete cosine transform (DCT)). , Integer transform, wavelet transform, or other conceptually similar transform) may include coefficients in the transform domain. The residual data may correspond to the pixel difference between the pixels of the unencoded picture and the predictor corresponding to the PU. The video encoder 20 can generate a TU containing the residual data of the CU and then convert the TU to generate a conversion factor of the CU.

After performing some conversion to generate the conversion factor, the video encoder 20 can quantize the conversion factor. Quantization refers to the process of quantizing the coefficients, for example to reduce the amount of data used to represent the coefficients and to perform further compression. The quantization process can reduce the bit depth associated with some or all of the coefficients. For example, during quantization, the n-bit value can be reduced to an m-bit value by rounding, where n is greater than m.

The JEM model further improves the video-picture coding structure. Specifically, a block coding structure called a "quad tree plus binary tree (QTBT) structure" will be introduced. Without using concepts like CU, PU, TU in HEVC, the QTBT structure supports a more flexible CU split shape. The CU may be square or rectangular. Quadtree partitioning is first performed on the CTU, and binary partitioning is further performed on the leaf node of the quadtree. In addition, there are two binary partitioning modes: symmetric horizontal partitioning and symmetric vertical partitioning. The binary tree leaf node is referred to as the CU. The CU in the JEM model cannot be further subdivided between prediction and transformation. In other words, the CU, PU, and TU in the JEM model have the same block size. In the existing JEM model, the maximum CTU size is 256 x 256 Luma pixels.

In some feasible implementations, the video encoder 20 scans the quantized conversion factors in a predetermined scan order to produce a serialized vector that can be entropy-coded. Can be done. In other feasible implementations, the video encoder 20 can perform adaptive scanning. After scanning the quantized transformation coefficients to form a one-dimensional vector, the video encoder 20 is subjected to context-adaptive variable-length coding (CAVLC) method, context-based adaptive binary. arithmetic coding (context-based adaptive binary arithmetic coding , CABAC) method, the probability interval partitioning entropy (probability interval partitioning entropy, PIPE) coding method, or another based on the entropy coding method, entropy sign-one-dimensional vector Can be transformed into. The video encoder 20 can further encode the syntax elements associated with the encoded video data, so that the video decoder 30 decodes the video data.

To perform CABAC, the video encoder 20 can assign a context in the context model to a symbol to be transmitted. The context can be associated with whether the adjacent value of the symbol is non-zero. To perform CAVLC, the video encoder 20 can select the variable length code of the symbol to be transmitted. Codewords (variable-length coding, VLC) in variable length coding are configured such that shorter codes correspond to more likely symbols and longer codes correspond to less likely symbols. Is possible. In this way, using VLC can reduce the bit rate as compared to using codewords of equal length for all transmitted symbols. Probabilities in CABAC can be determined based on the context assigned to the symbol.

In this embodiment of the present application, the video encoder can perform inter-prediction to reduce temporal redundancy between pictures. As mentioned above, the CU can have one or more prediction unit PUs according to different video compression coding standards. In other words, multiple PUs may belong to one CU, or the PUs and CUs have the same size. As used herein, if the CU and PU have the same size, the partitioning mode of the CU either performs no partitioning or divides the CU into one PU, where the PUs are uniform for illustration purposes. Used for. When the video encoder performs inter-prediction, the video encoder can signal motion information about the PU to the video decoder. For example, motion information about a PU may include a reference picture index, a motion vector, and a predictive direction identifier. The motion vector can indicate the displacement between the picture block of the PU (also called a video block, pixel block, pixel set, etc.) and the reference block of the PU. The reference block of the PU may be part of a reference picture similar to the picture block of the PU. The reference block may be placed within the reference picture indicated by the reference picture index and the predictive direction identifier.

To reduce the amount of coded bits needed to represent motion information about the PU, the video encoder follows a merge prediction mode or advanced motion vector prediction (AMVP) mode. A list of candidate predicted motion vectors (MVs) for each PU can be generated. Each of the candidate predicted motion vectors in the list of candidate predicted motion vectors for the PU can indicate motion information. The motion information represented by some candidate predicted motion vectors in the list of candidate predicted motion vectors may be based on motion information for other PUs. Candidate prediction if the candidate's predicted motion vector indicates motion information for one of a particular spatial candidate's predicted motion vector position or a particular temporal candidate's predicted motion vector position. The resulting motion vector may be referred to herein as the predicted motion vector of the "original" candidate. For example, in the merge mode, which is also referred to as the merge prediction mode in the present application, there may be a predicted motion vector position of five original space candidates and a predicted motion vector position of one original time candidate. In some examples, the video encoder may combine several motion vectors from different original candidate predicted motion vectors, modify the original candidate's predicted motion vector, or predict the candidate. By inserting only the zero motion vector as the motion vector, the predicted motion vector of additional candidates can be generated. The predicted motion vectors of these additional candidates are not considered to be the predicted motion vectors of the original candidate and may be referred to herein as the manually generated predicted motion vectors of the candidates.

Techniques in the present application typically include techniques for generating a list of predicted motion vectors for candidates in a video encoder and techniques for generating a list of predicted motion vectors for the same candidate in a video decoder. Video encoders and video decoders can generate a list of predicted motion vectors for the same candidate by implementing the same technique for constructing a list of predicted motion vectors for the same candidate. For example, a video encoder and a video decoder can construct a list with the same amount of candidate predicted motion vectors (eg, 5 candidate predicted motion vectors). Video encoders and video decoders first consider the predicted motion vectors of spatial candidates (eg, adjacent blocks in the same picture), and then (eg, predicted candidates in different pictures). Considering the predicted motion vectors of temporal candidates (such as motion vectors) and finally considering the predicted motion vectors of manually generated candidates, the required amount of candidate predictions. This can be done until the motion vector is added to the list. According to the technique of the present application, there is to remove the iterative candidate's predicted motion vector from the candidate's list of predicted motion vectors while constructing the candidate's list of predicted motion vectors. A pruning process may be performed on the predicted motion vector of the type candidate, and the pruning process is performed on the predicted motion vector of the other type candidate in order to reduce the complexity of the decoder. It does not have to be. For example, for a set of predicted motion vectors of spatial candidates and a set of predicted motion vectors of temporal candidates, the pruning process combines the predicted motion vectors of the candidates with iterative motion information. It can be performed to remove from the list of candidate predicted motion vectors. However, the manually generated candidate predicted motion vector is added to the list of candidate predicted motion vectors if the pruning process is not performed on the manually generated candidate predicted motion vector. It is possible.

After generating a list of candidate predicted motion vectors for the PU of the CU, the video encoder selects the candidate predicted motion vector from the list of candidate predicted motion vectors and the candidate predicted motion vector. The motion vector index can be output as a bit stream. The predicted motion vector of the selected candidate may be a candidate predicted motion vector for generating a motion vector that best matches the predicted unit of the decoded target PU. The candidate's predicted motion vector index can indicate the position of the selected candidate's predicted motion vector in the list of candidate's predicted motion vectors. The video encoder may further generate a predictive picture block for the PU based on the reference block indicated by the motion information about the PU. The motion information about the PU may be determined based on the motion information indicated by the predicted motion vector of the selected candidate. For example, in merge mode, the motion information of the PU may be the same as the motion information indicated by the predicted motion vector of the selected candidate. In the advanced motion vector prediction mode, the motion information for the PU can be determined based on the motion vector difference for the PU and the motion information indicated by the predicted motion vector of the selected candidate. .. The video encoder can generate one or more residual picture blocks for the CU based on the predicted picture blocks for the PU of the CU and the original picture blocks for the CU. The video encoder can then encode one or more residual picture blocks and output one or more residual picture blocks in a bitstream.

The bitstream can contain data that identifies the predicted motion vectors of the selected candidates in the list of predicted motion vectors of the candidates for the PU. The video decoder can determine the motion information of the PU based on the motion information indicated by the predicted motion vector of the selected candidate in the candidate predicted motion vector list for the PU. The video decoder can identify one or more reference blocks for the PU based on the motion information of the PU. After identifying one or more reference blocks for the PU, the video decoder may generate a predictive picture block for the PU based on the one or more reference blocks for the PU. The video decoder can reconstruct the picture block for the CU based on the predicted picture block for the PU of the CU and one or more residual picture blocks for the CU.

For the sake of brevity, the location or picture block can be described herein as having various spatial relationships with the CU or PU. The description can be described as follows: The location or picture block has various spatial relationships with the picture block associated with the CU or PU. Additionally, in the present application, the PU currently being decoded by the video decoder may be referred to as the PU to be processed, or may be referred to as the picture block currently being processed. In the present application, the CU currently decoded by the video decoder may be referred to as the CU to be processed. In the present application, the picture currently decoded by the video decoder may be referred to as the current picture. It should be understood that the present application is also applicable if the PU and CU have the same size, or if the PU is a CU. PU is used uniformly for illustration purposes.

As briefly described above, the video encoder 20 can generate predictive picture blocks and motion information for the PU of the CU by inter-prediction. In many examples, the motion information of a given PU is of one or more adjacent PUs (ie, a PU in which a picture block is spatially or temporally adjacent to a picture block of a given PU). It may be the same as or similar to the motion information. Since the adjacent PUs often have similar motion information, the video encoder 20 can encode the motion information of a given PU based on the motion information of the adjacent PUs. Encoding the motion information of a given PU based on the motion information of adjacent PUs reduces the amount of encoded bits required in the bitstream to indicate the motion information of a given PU. Can be done.

The video encoder 20 can encode the motion information of a given PU based on the motion information of the adjacent PUs in various ways. For example, the video encoder 20 can indicate that the motion information for a given PU is the same as the motion information for an adjacent PU. In the present application, the merge mode may be used to indicate that the motion information for a given PU is the same as the motion information for an adjacent PU, or it may be derived from the motion information for an adjacent PU. good. In another feasible implementation, the video encoder 20 is capable of calculating a motion vector difference (MVD) for a given PU, and the MVD is an initial motion vector for a given PU. And the difference between the final motion vector for a given PU. The video encoder 20 can include the MVD in place of the motion vector of a given PU in the motion information of a given PU. In a bitstream, the amount of coded bits needed to represent an MVD is less than the amount of coded bits needed to represent a motion vector for a given PU. The present application uses an advanced motion vector prediction mode to show that motion information for a given PU is signaled to the decoder by using an MVD and an index value to identify a candidate motion vector. be able to.

In order to signal the motion information of a given PU to the decoder in merge mode or AMVP mode, the video encoder 20 can generate a list of candidate predicted motion vectors for a given PU. The list of candidate predicted motion vectors may include one or more candidate predicted motion vectors. Each candidate's predicted motion vector in the list of candidate predicted motion vectors for a given PU can specify motion information. The motion information represented by each candidate's predicted motion vector may include a motion vector, a reference picture index, and a predicted direction identifier. The predicted motion vectors of the candidates in the list of predicted motion vectors of the candidates may contain the predicted motion vectors of the "original" candidates, and the predicted motion vectors of each "original" candidate , Shows motion information of one of the predicted motion vector positions of a particular candidate in a PU different from a given PU.

After generating a list of candidate predicted motion vectors for the PU, the video encoder 20 picks one of the candidate predicted motion vectors from the list of candidate predicted motion vectors for the PU. You can choose. For example, the video encoder can compare each candidate's predicted motion vector to the decoded PU and select a candidate's predicted motion vector with the desired rate distortion cost. The video encoder 20 can output a candidate predicted motion vector index for the PU. The candidate's predicted motion vector index can identify the position of the selected candidate's predicted motion vector in the candidate's predicted motion vector list.

Further, the video encoder 20 may generate a predicted picture block of the PU based on the reference block indicated by the motion information of the PU. The motion information for the PU may be determined based on the motion information indicated by the predicted motion vector of the candidate selected in the list of predicted motion vectors for the candidate for the PU. For example, in merge mode, the motion information of the PU may be the same as the motion information indicated by the predicted motion vector of the selected candidate. In the AMVP mode, the motion information of the PU may be determined based on the motion vector difference of the PU and the motion information indicated by the predicted motion vector of the selected candidate. As mentioned above, the video encoder 20 can process the predicted picture block of the PU.

When the video decoder 30 receives the bitstream, the video decoder 30 can generate a list of candidate predicted motion vectors for each PU of the CU. The list of candidate predicted motion vectors generated by the video decoder 30 for the PU may be the same as the list of candidate predicted motion vectors generated by the video encoder 20 for the PU. The syntax element obtained by analyzing the bitstream can indicate the position of the predicted motion vector of the selected candidate in the list of predicted motion vectors of the candidate for the PU. After generating a list of candidate predicted motion vectors for the PU, the video decoder 30 may also generate a PU predictive picture block based on one or more reference blocks indicated by the PU motion information. good. The video decoder 30 can determine the motion information of the PU based on the motion information indicated by the predicted motion vector of the candidate selected in the list of predicted motion vectors of the candidates for the PU. The video decoder 30 can reconstruct the picture block of the CU based on the predicted picture block of the PU and the residual picture block of the CU.

In a feasible implementation, the bits for constructing a list of candidate predicted motion vectors and obtaining the position of the selected candidate's predicted motion vector in the candidate's list of predicted motion vectors on the decoder. It should be understood that the analysis of the stream is independent of each other and may be performed in any order or in parallel.

In another feasible implementation, on the decoder, the position of the predicted motion vector of the candidate selected in the list of predicted motion vectors of the candidate is first obtained by analyzing the bit stream and then the candidate. A list of predicted motion vectors of is constructed based on the positions obtained by the analysis. In this implementation, it is not necessary to build a list of predicted motion vectors for all candidates, only the list of predicted motion vectors for candidates at the position obtained by the analysis is the candidate at that position. It needs to be constructed to determine the predicted motion vector. For example, by analyzing the bit stream, the predicted motion vector of the selected candidate is the predicted motion vector of the candidate having an index of 3 in the list of predicted motion vectors of the candidate and having an index of 3. It can be seen that in order to determine the predicted motion vector of the candidate, it is necessary to construct only the list of the predicted motion vector of the candidate from index 0 to index 3. This can reduce complexity and improve decoding efficiency.

FIG. 2 is a schematic block diagram of the video encoder 20 according to the embodiment of the present application. The video encoder 20 can perform intra-decoding and inter-decoding on the video blocks in the video slice. Intra-decoding relies on spatial prediction to reduce or eliminate spatial redundancy of video within a given video frame or picture. Interdecoding relies on temporal prediction to reduce or eliminate temporal redundancy of video in adjacent frames or pictures of a video sequence. The intra mode (I mode) may be any one of several spatial compression modes. The inter-mode, such as one-way prediction mode (P mode) or two-way prediction mode (B mode), may be any one of several temporal compression modes.

In a feasible implementation of FIG. 2, the video encoder 20 includes a partitioning unit 35, a prediction unit 41, a reference picture storage 64, an adder 50, a conversion unit 52, a quantization unit 54, and an entropy coding unit 56. include.

The reference picture storage 64 provides a target storage space for storing the MVD, i.e., to store the difference between the final motion vector and the initial motion vector for the PU to be processed. , Further configured. In some embodiments, the reference picture storage 64 can further store the final motion vector of the PU to be processed. In this embodiment of the present application, the MVD stored in the reference picture storage 64 may be used in another PU coding process. For example, in the decryption process in the merge mode, the unupdated initial motion vector and the updated final motion vector are stored separately at different times. Therefore, if the PU to be processed needs to obtain the predicted motion vector for another PU, under certain conditions, the PU to be processed will get the unupdated initial motion vector of the other PU. , It can be used directly as a motion vector of another PU , and it is not necessary to obtain a motion vector of another PU after the motion vector of another PU is updated.

Note that the target storage space may be provided by memory other than the reference picture storage 64, eg, newly added memory. This is not specifically limited in this embodiment of the present application.

In a related technique, when the video encoder 20 operates in merge mode, the contents of the reference picture storage 64 are empty or zero vector, i.e., the video encoder 20 is in merge mode. When operating, the target storage space is not used during encoding, and in the encoding process, the predicted motion vector of the other PUs obtained by the PU being processed is the updated final of the other PUs. Note that it is a motion vector. Therefore, the PU to be processed can obtain the predicted motion vector for the other PU only after the predicted motion vector of the other PU has been updated, and as a result, the coding during the coding. There is a delay in conversion. However, if the video encoder 20 operates in non-merge mode, the reference picture storage 64 may be used. For example, in AMVP mode, the target storage space is used to store the motion vector residuals, and the video encoder 20 encodes the motion vector residuals and, as a result, after decoding the current PU. The video decoder 30 acquires the motion vector residual of the PU to be processed, and is the final of the PU to be processed based on the motion vector residual of the PU to be processed and the initial motion vector of the PU to be processed. Motion vector can be obtained.

The prediction unit 41 includes a motion estimation unit 42, a motion compensation unit 44, and an intra prediction unit 46. For video block reconstruction, the video encoder 20 further includes an inverse quantization unit 58, an inverse transformation unit 60, and an adder 62. The video encoder 20 may further include a deblocking filter (not shown in FIG. 2) to perform filtering on the block boundaries and remove blocking artifacts from the reconstructed video. If necessary, the deblocking filter typically performs filtering on the output of adder 62. In addition to the deblocking filter, additional loop filters (inside or after the loop) may be used.

As shown in FIG. 2, the video encoder 20 receives the video data and the partitioning unit 35 divides the data into video blocks. Such partitioning may further include partitioning into slices, picture blocks, or other larger units and video block partitioning based on the LCU and CU quadtree structure. For example, the video encoder 20 is a component for encoding a video block in a video slice to be encoded. Usually, one slice can be divided into multiple video blocks (and can be divided into a set of video blocks called picture blocks).

The prediction unit 41 may have a plurality of possible decoding modes, eg, a plurality, for the current video block, based on the coding quality and the cost calculation result (eg, rate distortion cost, RDcost ). You can select one of the intra-decoding modes or one of the plurality of inter-decoding modes. The prediction unit 41 provides the acquired intra-decoding block or inter-decoding block to the adder 50 to generate residual block data, and the obtained intra-decoding block or inter-decoding block is added to the adder 62. The encoded block can be reconstructed and the reconstructed encoded block can be used as a reference picture.

The motion estimation unit 42 and motion compensation unit 44 in the prediction unit 41 perform interpredictive decoding for the current video block for one or more predictive blocks of one or more reference pictures to perform temporal compression. .. The motion estimation unit 42 may be configured to determine the inter-prediction mode of the video slice based on the preset mode of the video sequence. In preset mode, the video slices in the sequence can be designated as P slices, B slices, GPB slices. The motion estimation unit 42 and the motion compensation unit 44 may be highly integrated, but are described separately to illustrate the concept. The motion estimation performed by the motion estimation unit 42 is a process of generating a motion vector for estimating a video block. For example, the motion vector can indicate the displacement of the PU of the current video frame or video block of the picture with respect to the predicted block of the reference picture.

A predictive block is a block of PU that is found to match closely to the video block to be decoded based on the pixel difference, and the pixel difference is the sum of absolute difference s ,. SAD), sum (sum of squared difference s of the square difference, SSD), or can be determined by using other difference metrics. In some feasible implementations, the video encoder 20 can calculate the value of the sub-integer pixel position of the reference picture stored in the reference picture storage 64. For example, the video encoder 20 can interpolate a value at a 1/4 pixel position, a 1/8 pixel position, or another fractional pixel position of a reference picture. Therefore, the motion estimation unit 42 can execute a motion search for all pixel positions and fractional pixel positions and output a motion vector with fractional pixel accuracy.

The motion estimation unit 42 calculates the motion vector for the PU of the video block in the interdecoded slice by comparing the position of the PU with the position of the predicted block of the reference picture. The reference picture can be selected from the first reference picture list (list 0) or the second reference picture list (list 1). Each list is used to identify one or more reference pictures stored in the reference picture storage 64. The motion estimation unit 42 sends the calculated motion vector to the entropy coding unit 56 and the motion compensation unit 44.

Motion compensation performed by motion compensation unit 44 can include extraction or generation of predictive blocks based on motion vectors determined by motion estimation and can perform interpolation with sub-pixel level accuracy. It is possible. After receiving the motion vector for the PU of the current video block, the motion compensation unit 44 can locate the predictive block that the motion vector points to in one reference picture list. The video encoder 20 subtracts the pixel value of the predicted block from the pixel value of the current video block being decoded to obtain the residual video block and obtains the pixel difference. The pixel difference constitutes the residual data of the block and may contain both the Luma difference component and the Chroma difference component. The adder 50 is one or more components that perform a subtraction operation. The motion compensation unit 44 can further generate the video blocks and the syntax elements associated with the video slices, so that the video decoder 30 decodes the video blocks in the video slices.

When the PU is located in the B slice, the picture containing the PU can be associated with two reference picture lists referred to as "list 0" and "list 1". In some feasible implementations, a picture containing a B slice may be associated with a combination of lists in Listing 0 and Listing 1.

Further, when the PU is located in the B slice, the motion estimation unit 42 can execute one-way prediction or two-way prediction for the PU. In some feasible implementations, bidirectional prediction is a prediction performed separately based on the pictures in reference picture list 0 and reference picture list 1. In some other feasible implementation, bidirectional prediction is a prediction that is performed separately based on the reconstructed future frame and the reconstructed past frame for the current frame in display order. be. Further, when the motion estimation unit 42 executes one-way prediction for the PU, the motion estimation unit 42 searches the reference picture in the list 0 or the list 1 in order to search for the reference block for the PU. Can be done. Then, the motion estimation unit 42 can generate a reference index indicating a reference picture including a reference block in list 0 or list 1, and a motion vector indicating a spatial displacement between the PU and the reference block. The motion estimation unit 42 can output a reference index, a prediction direction identifier, and a motion vector as motion information of the PU. The predictive direction identifier may indicate that the reference index indicates a reference picture in Listing 0 or Listing 1. The motion compensation unit 44 can generate a predicted picture block of the PU based on the reference block indicated by the motion information of the PU.

When the motion estimation unit 42 makes bidirectional predictions for the PU, the motion estimation unit 42 searches for a reference picture in list 0 for a reference block for the PU, and further for another reference block for the PU. The reference picture in the list 1 can be searched. Then, the motion estimation unit 42 can generate a reference index indicating a reference picture including the reference blocks of List 0 and Listing 1, and a motion vector indicating a spatial displacement between the reference block and the PU. Further, the motion estimation unit 42 can output the reference index and the motion vector of the PU as the motion information of the PU. The motion compensation unit 44 can generate a predicted picture block of the PU based on the reference block indicated by the motion information of the PU.

In some feasible implementations, motion estimation unit 42 does not output a complete set of PU motion information to entropy coding unit 56. Instead, the motion estimation unit 42 may signal PU motion information regarding motion information of another PU. For example, the motion estimation unit 42 can determine that the motion information of the PU is similar to the motion information of the adjacent PU. In this implementation, the motion estimation unit 42 can indicate an indicator value in the syntax structure related to the PU, and the indicator value is that the motion information of the PU is the same as the motion information of the adjacent PU. , Or the video decoder 30 may be derived from motion information of adjacent PUs. In another implementation, the motion estimation unit 42 can discriminate between the predicted motion vector of a candidate associated with an adjacent PU and the motion vector difference (MVD) in the PU-related syntax structure. can. The MVD indicates the difference between the motion vector of the PU and the predicted motion vector of the indicated candidate associated with the adjacent PU. The video decoder 30 can determine the motion vector of the PU by using the predicted motion vector and MVD of the indicated candidate.

As mentioned above, the prediction unit 41 can generate a list of candidate predicted motion vectors for each PU of the CU. The list of predicted motion vectors for one or more candidates is one or more additions derived from the predicted motion vectors for one or more original candidates and the predicted motion vectors for one or more original candidates. Can include candidate predicted motion vectors.

In this embodiment of the present application, when setting a list of candidate predicted motion vectors in merge mode, the prediction unit 41 adds the candidate predicted motion vector list to the reference PU of the PU to be processed. The motion vector can be stored and the motion vector of the reference PU may be the unupdated initial motion vector of the reference PU or the updated final motion vector of the reference PU. It should be noted that it is good. The updated final motion vector for the reference PU may be obtained based on the motion vector stored in the target storage space. If the target storage space stores the MVD, the final motion vector of the reference PU may be obtained by adding the initial motion vector of the reference PU to the MVD stored in the target storage space. It is possible. If the target storage space stores the final motion vector for the reference PU, the final motion vector for the reference PU can be obtained directly from the target storage space. The motion vector of the reference PU can be stored in the list of candidate predicted motion vectors in the following way: if the reference PU and the current PU are in the same CTB or CTB row range. the initial motion vectors that have not been updated reference unit is either stored in the list of predicted motion vector candidate, or if the reference PU and the current PU is not in the line range of the same CTB or CTB Stores the updated final motion vector of the reference unit in the list of candidate predicted motion vectors.

The intra prediction unit 46 of the prediction unit 41 performs intra prediction decoding for the current video block for one or more adjacent blocks in the same picture or slice as the current block to be decoded. Can be performed and provide spatial compression. Therefore, as an alternative to the inter-prediction performed by the motion estimation unit 42 and the motion compensation unit 44 (as described above), the intra-prediction unit 46 can perform the intra-prediction on the current block. Specifically, the intra prediction unit 46 can determine the intra prediction mode for encoding the current block. In some feasible implementations, the intra-prediction unit 46 (eg) uses various intra-prediction modes to encode the current block during separate coding traversals, and the intra-prediction unit 46 (or). The mode selection unit 40) in some feasible implementations can select the appropriate intra-prediction mode from the test mode.

After the prediction unit 41 generates a prediction block of the current video block by inter-prediction or intra-prediction, the video encoder 20 subtracts the prediction block from the current video block to obtain the residual video block. do. The residual video data in the residual block is contained in one or more TUs and can be applied to the conversion unit 52. The transform unit 52 performs a transform such as a discrete cosine transform (DCT) or a conceptually similar transform (eg, a discrete sine transform (DST)) to perform a residual video. Convert the data to the residual transform coefficient. The conversion unit 52 can convert the residual video data from the pixel domain to the conversion domain (eg, the frequency domain).

The conversion unit 52 can send the acquired conversion coefficient to the quantization unit 54. The quantization unit 54 quantizes the conversion factor to further reduce the bit rate. The quantization process can reduce the bit depth associated with some or all of the coefficients. The degree of quantization can be modified by adjusting the quantization parameters. In some feasible implementations, the quantized unit 54 can then scan the matrix containing the quantized transformation coefficients. Alternatively, the entropy coding unit 56 may perform the scan.

After quantization, the entropy coding unit 56 can entropy code the quantized conversion factors. For example, the entropy coding unit 56 may include context-adaptive variable-length coding (CAVLC), context-adaptive binary arithmetic coding (CABAC), probability interval partitioning entropy (PIPE) coding, or other entropy coding methods or techniques. Can be executed. The entropy coding unit 56 can further entropy code the motion vector of the current video slice being decoded and another syntax element. After being entropy-coded by the entropy coding unit 56, the encoded bitstream may be transmitted to the video decoder 30 or recorded for subsequent transmission or retrieval by the video decoder 30. May be good.

The entropy coding unit 56 can encode information indicating the selected intra prediction mode according to the technique of the present application. The video encoder 20 is also referred to as a plurality of intra-prediction mode index tables and a plurality of modified intra-prediction mode index tables (also referred to as codeword mapping tables) used for each of the contexts. ) Can be included in the transmitted bitstream configuration data, such as definitions of coding contexts for various blocks, indications for most probable mode (MPM), intra-predictive mode index tables, and It can contain a modified intra-prediction mode index table.

The inverse quantization unit 58 and the inverse transformation unit 60 perform inverse quantization and inverse transformation, respectively, to reconstruct the residual block in the pixel domain, which is subsequently used as the reference block of the reference picture. The motion compensation unit 44 can calculate the reference block by adding the residual block and the prediction block of one reference picture in one reference picture list. Further, the motion compensation unit 44 may apply one or more interpolation filters to the reconstructed residual block to calculate a sub-integer pixel value for motion estimation. The adder 62 adds the reconstructed residual block and the motion compensated prediction block generated by the motion compensation unit 44 to generate a reference block stored in the reference picture storage 64. The reference block may be used by the motion estimation unit 42 and the motion compensation unit 44 as a reference block for performing interprediction with respect to a block in a subsequent video frame or picture.

FIG. 3 is a schematic block diagram of the video decoder 30 according to the embodiment of the present application. In a feasible implementation of FIG. 3, the video decoder 30 includes an entropy decoding unit 80, a prediction unit 81, an inverse quantization unit 86, an inverse conversion unit 88, an adder 90, and a reference picture storage 92.

The reference picture storage 92 can be configured to provide a target storage space. The target storage space is used to store the MVD, i.e., to store the difference between the final motion vector and the initial motion vector. In some embodiments, the target storage space is used to store the final motion vector. In this embodiment of the present application, the updated final motion vector, or MVD stored in the target storage space, can be used in another PU coding process. For example, in the decryption process in merge mode, the unupdated initial motion vector and the updated final motion vector are stored separately at different times. Therefore, if the PU to be processed needs to obtain the predicted motion vector for another PU, under certain conditions, the PU to be processed will get the unupdated initial motion vector of the other PU. , It can be used directly as a motion vector of another PU, and it is not necessary to obtain the motion vector of another block after the motion vector of another PU is updated.

It should be noted that the target storage space may be provided by memory other than the reference picture storage 92, eg, newly added memory. This is not specifically limited to the embodiments of the present application.

In a related technique, when the video decoder 30 operates in merge mode, the contents of the target storage space are empty or zero vector, that is, the video decoder 30 operates in merge mode. If so, the target storage space is not used during decoding, and during decoding, the predicted motion vectors of the other PUs obtained by the PU being processed are updated by the other PUs. It should be noted that it is the final motion vector. Therefore, the PU to be processed can obtain the predicted motion vector for the other PU only after the predicted motion vector for the other PU has been updated, and as a result, decode during decoding. There will be a delay in conversion. However, if the video decoder 30 operates in non-merge mode, the target storage space may be used. For example, in AMVP mode, the video decoder 30 can obtain the MVD for the PU to be processed by decoding, so that the video decoder 30 has the initial motion vector and the PU to be processed. The final motion vector for the PU to be processed can be obtained based on the MVD.

The prediction unit 81 includes a motion compensation unit 82 and an intra prediction unit 84. In some feasible implementations, the video decoder 30 can perform an exemplary decoding process that is the reverse of the coding process described for the video encoder 20 of FIG.

During decoding, the video decoder 30 receives from the video encoder 20 an encoded video bitstream representing a video block of encoded video slices and associated syntax elements. The entropy decoding unit 80 of the video decoder 30 decodes the bitstream to generate quantized coefficients, motion vectors, and other syntax elements. The entropy decoding unit 80 transfers the motion vector and other syntax elements to the prediction unit 81. The video decoder 30 can receive syntax elements at the video slice level and / or the video block level.

If the video slice is decoded into an intra-decrypted (I) slice, the intra-prediction unit 84 of the prediction unit 81 is decoded prior to the signaled intra-prediction mode and the current frame or picture. Predictive data for the video block of the current video slice can be generated based on the data of the block.

When the video picture is decoded into an inter-decoded slice (eg, a B-slice, a P-slice, or a GPB slice), the motion compensation unit 82 of the prediction unit 81 receives the motion received from the entropy-decoding unit 80. Generates a predictive block of the video block of the current video picture based on vectors and other syntax elements. The prediction block may be generated from one reference picture in one reference picture list. The video decoder 30 can use the default configuration technique to construct the reference picture list (List 0 and List 1) based on the reference picture stored in the reference picture storage 92.

The motion compensation unit 82 determines the prediction information of the video block of the current video slice by analyzing the motion vector and other syntax elements, and uses the prediction information to decode the video block. Generate a prediction block for. For example, motion compensation unit 82 uses some of the received syntax elements to decode a video block of a video slice in a predictive (eg, intra-predictive or inter-predictive) mode, inter-predictive slice type. (For example, B slice, P slice, or GPB slice) Configuration information of one or more reference picture lists of the slice, motion vector of each inter-encoded video block of the slice, inter-decoding of each of the slices. Determines the interpredicted status of the video block that has been made, as well as other information for decoding the video block in the current video slice.

The motion compensation unit 82 may further perform interpolation using an interpolation filter. The motion compensation unit 82 can calculate the interpolated value of the sub-integer pixels of the reference block, for example, using the interpolation filter used by the video encoder 20 during the coding of the video block. In the present application, the motion compensation unit 82 can determine the interpolation filter used by the video encoder 20 based on the received syntax element, and use the interpolation filter to generate a prediction block.

If the PU is encoded by inter-prediction, the motion compensation unit 82 can generate a list of candidate predicted motion vectors for the PU. The bitstream may contain data for identifying the position of the predicted motion vector of the selected candidate in the list of predicted motion vectors of the candidate for the PU. After generating the predicted motion vector of the candidate for the PU, the motion compensation unit 82 can generate the predicted picture block for the PU based on one or more reference blocks indicated by the motion information for the PU. The reference block of the PU may be located in a temporal picture different from the temporal picture of the PU. The motion compensation unit 82 can determine the motion information of the PU based on the motion information selected in the list of predicted motion vectors of the candidates for the PU.

In an embodiment of the present application, the motion compensation unit 82 predicts a candidate motion vector with respect to a reference PU of the PU to be processed when setting a list of candidate predicted motion vectors in merge mode. The motion vector may be stored in the list of motion vectors, and the predicted motion vector of the reference PU may be the initial motion vector of the reference PU that has not been updated, or the final motion of the reference PU that has been updated. It should be noted that it may be a vector. The updated final motion vector of the reference PU may be obtained based on the motion vector for the reference PU stored in the target storage space. For example, if the target storage space stores the MVD, the final motion vector for the reference PU is by adding the initial motion vector for the reference PU and the MVD stored in the target storage space. , Can be obtained. If the target storage space stores the final motion vector for the reference PU, the final motion vector for the reference PU can be obtained directly from the target storage space. .. The predicted motion vector for the reference PU may be stored in the list of candidate predicted motion vectors in the following way: the reference PU and the current PU are in the same CTB or CTB row range. If the initial motion vectors that have not been updated for the reference unit is stored in the list of predicted motion vector candidates; or if the reference PU and the current PU is not in the line range of the same CTB or CTB , The updated final motion vector for the reference unit is stored in the list of candidate predicted motion vectors.

The dequantization unit 86 performs dequantization (eg, dequantization) on the quantization conversion coefficients provided in the bitstream and decoded by the entropy decoding unit 80. The dequantization process determines the degree of quantization based on the quantization parameters calculated by the video encoder 20 for each video block in the video slice, as well as the degree of dequantization to be applied. It is possible to include that. The inverse transformation unit 88 performs an inverse transformation on the transformation coefficients (eg, an inverse DCT, an inverse integer transformation, or a geometrically similar inverse transformation process) to generate a residual block in the pixel domain.

After the motion compensation unit 82 has generated a predictive block of the current video block based on the motion vector and other syntax elements, the video decoder 30 has a residual block from the inverse conversion unit 88 and motion compensation. Performs an addition operation with the corresponding predictive block generated by the unit 82 to form the decoded video block. The adder 90 is one or more components that perform an addition operation. If desired, a deblocking filter may be used to perform filtering on the decoded blocks to remove blocking artifacts. Another loop filter (inside or after the decoding loop) may be used to smooth the pixels, or the video quality may be improved in another way. The decoded video block within a given frame or picture is then stored in reference picture storage 92. The reference picture storage 92 stores a reference picture used for subsequent motion compensation. The reference picture storage 92 also stores the decoded video that is subsequently presented in a display device such as the display device 32 of FIG.

As mentioned above, the techniques of the present application relate, for example, to inter-decoding. The techniques of the present application may be performed by any video decoder described herein, the video decoders being (eg) the video encoder 20 and the video decoder 30 shown and described in FIGS. 1 to 3. It should be understood to include. Specifically, in a feasible implementation, the prediction unit 41 described in FIG. 2 implements the specific techniques described below when inter-prediction is performed while encoding a block of video data. Can be executed. In another feasible implementation, the prediction unit 81 described in FIG. 3 performs the specific technique described below when inter-prediction is performed during decoding of a block of video data. Can be done. Thus, a reference to a general "video encoder" or "video decoder" may include a video encoder 20, a video decoder 30, or another video coding or coding unit.

FIG. 4 is a schematic block diagram of an inter-prediction module according to an embodiment of the present application. The inter-prediction module 121 can include, for example, a motion estimation unit 42 and a motion compensation unit 44. The relationship between PU and CU depends on the video compression coding standard. The inter-prediction module 121 can divide the current CU into PUs according to a plurality of partitioning patterns. For example, the inter-prediction module 121 can classify the current CU into PUs according to a 2Nx2N, 2NxN, Nx2N, NxN partitioning pattern. In another embodiment, the current CU is the current PU. Not limited to this.

The inter-prediction module 121 can perform integer motion estimation (IME) and then fractional motion estimation (FME) for each PU. When the inter-prediction module 121 performs an IME on the PU, the inter-prediction module 121 can search for one or more reference pictures for the reference block for the PU. After discovering the reference block for the PU, the inter-prediction module 121 can generate a motion vector indicating the spatial displacement between the PU and the reference block for the PU with integer accuracy. When the inter-prediction module 121 executes the FME on the PU, the inter-prediction module 121 can improve the motion vector generated by executing the IME on the PU. The motion vector generated by running FME on the PU may have sub-integer accuracy (eg, 1/2 pixel accuracy or 1/4 pixel accuracy). After generating the motion vector of the PU, the inter-prediction module 121 can generate the predicted picture block of the PU by using the motion vector of the PU.

In some feasible implementations in which the inter-prediction module 121 signals the PU motion information to the decoder in AMVP mode, the inter-prediction module 121 may generate a list of candidate predicted motion vectors for the PU. can. The list of predicted motion vectors of candidates is the predicted motion vector of one or more original candidates and the predicted motion vector of one or more additional candidates derived from the predicted motion vector of one or more original candidates. It may also include a motion vector. After generating a list of candidate predicted motion vectors for the PU, the inter-prediction module 121 selects the candidate predicted motion vectors from the list of candidate predicted motion vectors and for the PU. Motion vector differences (MVDs) can be generated. The MVD for the PU can indicate the difference between the motion vector represented by the predicted motion vector of the selected candidate and the motion vector generated for the PU by the IME and FME. In these feasible implementations, the inter-prediction module 121 indexes the candidate's predicted motion vectors to identify the position of the selected candidate's predicted motion vector in the list of candidate's predicted motion vectors. Can be output. The inter-prediction module 121 can further output the MVD of the PU. Hereinafter, a feasible implementation of the advanced motion vector prediction (AMVP) mode will be described in detail in FIG. 11 in the embodiment of the present application.

In addition to executing IME and FME on the PU to generate motion information on the PU, the inter-prediction module 121 may further perform a merge operation on the PU. When the inter-prediction module 121 performs a merge operation on the PU, the inter-prediction module 121 can generate a list of candidate predicted motion vectors for the PU. The list of candidate predicted motion vectors for PU is one or more original candidate predicted motion vectors and one or more additions derived from one or more original candidate predicted motion vectors. It may include a candidate's predicted motion vector. The predicted motion vector of one or more original candidates in the list of predicted motion vectors of the candidate includes the predicted motion vector of one or more spatial candidates and the predicted motion vector of the time candidate. Can be done. The predicted motion vector of the spatial candidate can indicate motion information for another PU in the current picture. The predicted motion vector of the time candidate may be based on the motion information of the corresponding PU in a picture different from the current picture. The predicted motion vector of a time candidate is sometimes referred to as a temporal motion vector predictor (TMVP).

After generating the list of candidate predicted motion vectors, the inter-prediction module 121 can select one candidate predicted motion vector from the list of candidate predicted motion vectors. Next, the inter-prediction module 121 can generate a predictive picture block of the PU based on the reference block indicated by the motion information of the PU. In merge mode, the motion information of the PU may be the same as the motion information indicated by the predicted motion vector of the selected candidate. FIG. 10 is a flowchart of an example of the merge mode.

After the IME and FME generate the predicted picture block of the PU and the merge operation to generate the predicted picture block of the PU, the inter-prediction module 121 may perform the predicted picture block or the predicted picture block generated by performing the FME operation. You can select the predicted picture blocks generated by performing the merge operation. In some feasible implementations, the inter-prediction module 121 has a rate distortion cost between the predicted picture block generated by performing the FME operation and the predicted picture block generated by performing the merge operation. By analyzing the, the predicted picture block of the PU can be selected.

After the inter-prediction module 121 selects the predicted picture block of the PU generated by partitioning the current CU according to each partitioning pattern (in some implementations, the coding tree unit CTU divides it into CUs). After that, the CU is not further subdivided into smaller PUs, in which case the PU is equivalent to the CU), the inter-prediction module 121 can select the partitioning pattern for the current CU. .. In some implementations, the inter-prediction module 121 analyzes the current rate distortion cost of the selected predicted picture block of the PU generated by partitioning the current CU according to each partitioning pattern. You can select the partitioning pattern for the CU. The inter-prediction module 121 can output the prediction picture block associated with the PU belonging to the selected partitioning pattern to the residual generation module 102. The inter-prediction module 121 can output the syntax element of the motion information of the PU belonging to the selected partitioning pattern to the entropy coding module 116.

In the schematic shown in FIG. 4, the inter-prediction module 121 includes IME modules 180A to 180N (collectively referred to as “IME module 180”), FME modules 182A to 182N (collectively referred to as “FME module 182”), and modules 184A to 184N. (Collectively referred to as "merge module 184"), PU pattern decision modules 186A to 186N (collectively referred to as "PU pattern decision module 186"), and CU pattern decision module 188 (CTU vs. CU pattern decision process). May be further executed).

The IME module 180, FME module 182, and merge module 184 can perform IME operations, FME operations, and merge operations on the PU of the current CU, respectively. In the schematic shown in FIG. 4, the inter-prediction module 121 is described as including a separate IME module 180, a separate FME module 182, and a separate merge module 184 for each PU in each partitioning pattern of the CU. Will be done. In another feasible implementation, the inter-prediction module 121 does not include a separate IME module 180, a separate FME module 182, or a separate merge module 184 for each PU in each partitioning pattern of the CU.

As shown in the schematic of FIG. 4, the IME module 180A, the FME module 182A, and the merge module 184A each have an IME for a PU generated by partitioning the CU according to a 2N × 2N partitioning pattern. Operations, FME operations, and merge operations can be performed. The PU mode decision module 186A can select one of the predictive picture blocks generated by the IME module 180A, the FME module 182A, and the merge module 184A.

The IME module 180B, FME module 182B, and merge module 184B each perform IME, FME, and merge operations on the left PU generated by partitioning the CU according to an N × 2N partitioning pattern. Can be executed. The PU mode decision module 186B can select one of the predictive picture blocks generated by the IME module 180B, the FME module 182B, and the merge module 184B.

The IME module 180C, FME module 182C, and merge module 184C each perform IME, FME, and merge operations on the right PU generated by partitioning the CU according to the N × 2N partitioning pattern. Can be executed. The PU mode decision module 186C can select one of the predictive picture blocks generated by the IME module 180C, the FME module 182C, and the merge module 184C.

The IME module 180N, the FME module 182N, and the merge module 184, respectively, perform an IME operation, an FME operation, and a merge operation on the lower right PU generated by partitioning the CU according to the N × N partitioning pattern. Can be executed. The PU mode decision module 186N can select one of the predictive picture blocks generated by the IME module 180N, the FME module 182N, and the merge module 184N.

The PU pattern decision module 186 selects a predictive picture block by analyzing the rate distortion cost of multiple possible predictive picture blocks and provides the optimal rate distortion cost for a given decoding scenario. -You can select a block. For example, for applications with limited bandwidth, the PU mode decision module 186 may prefer predictive picture blocks with increased compression ratios, and for other applications PU mode. The decision module 186 may prefer predictive picture blocks that improve the quality of the reconstructed video. After the PU pattern decision module 186 selects the predicted picture block for the PU of the current CU, the CU pattern decision module 188 selects the partitioning pattern for the current CU to the selected partitioning pattern. Outputs the predicted picture block and motion information for the PU to which it belongs.

FIG. 5 is a schematic diagram of an example of a coding unit and adjacent picture blocks associated with the coding unit according to an embodiment of the present application, with predicted movements of the CU 250 and exemplary candidates associated with the CU 250. It is a schematic diagram which shows the vector positions 252A to 252E. In the present application, the candidate's predicted motion vector positions 252A to 252E may be collectively referred to as the candidate's predicted motion vector position 252. The candidate's predicted motion vector position 252 represents the predicted motion vector of the spatial candidate in the same picture as the CU 250. The candidate's predicted motion vector position 252A is to the left of the CU250. The candidate's predicted motion vector position 252B is above the CU250. The candidate's predicted motion vector position 252C is in the upper right corner of the CU250. The candidate's predicted motion vector position 252D is at the bottom left of the CU250. The candidate's predicted motion vector position 252E is at the top left of the CU250. FIG. 8 shows an implementation example of how the inter-prediction module 121 and motion compensation module 162 can generate a list of candidate predicted motion vectors. Hereinafter, the implementation will be described in relation to the inter-prediction module 121. However, it should be understood that the motion compensation module 162 may implement the same technique and thus generate a list of predicted motion vectors of the same candidate. In this embodiment of the present application, the picture block in which the predicted motion vector position of the candidate is located is referred to as a reference block. Further, the reference block includes a spatial reference block, eg, a picture block in which 252A to 252E are located, and a time reference block, eg, a picture block in which a co-located block is located, or is co-located. Contains picture blocks that are spatially adjacent to each other.

FIG. 6 is an exemplary flow chart for constructing a list of candidate predicted motion vectors according to an embodiment of the present application. Although the technique of FIG. 6 is described on the basis of a list containing five candidate predicted motion vectors, the technique described herein may optionally be used with a list of different sizes. Each of the five candidate predicted motion vectors can have an index (eg, 0-4). The technique of FIG. 6 will be described on the basis of a typical video decoder. A typical video decoder may be, for example, a video encoder (eg, video encoder 20) or a video decoder (eg, video decoder 30).

In order to reconstruct the list of predicted motion vectors of the candidates according to the implementation of FIG. 6, the video decoder first considers four spatial candidates (602), and each spatial candidate has one predicted. Corresponding to the motion vector, the predicted motion vector of the four spatial candidates may include the candidate's predicted motion vector positions 252A, 252B, 252C, 252D. The predicted motion vectors of the four spatial candidates correspond to the motion information of the four PUs in the same picture as the current CU (eg, CU250). In other words, the predicted motion vectors of the four spatial candidates are the predicted motion vectors of the four PUs. In this embodiment of the present application, the video decoder must first determine the four PUs when acquiring the predicted motion vectors of the four spatial candidates. For example, the process by which the video decoder determines one of the four PUs may be as follows: if the PU and the current CU are in the same CTB or CTB row range: , The unupdated initial motion vector of the PU is used as the predicted motion vector of the spatial candidate corresponding to the PU, or the PU and the current CU are not in the same CTB or CTB row range. The updated final motion vector of the PU is used as the predicted motion vector of the spatial candidates corresponding to the PU. The updated final motion vector for the PU can be obtained from the target storage space where the PU is stored. In this case, the target storage space stores the updated final motion vector. In some embodiments, the target storage space stores the MVD. In this case, the updated final motion vector for the PU may instead be obtained based on the MVD stored in the target storage space and the unupdated initial motion vector for the PU. good.

The video decoder can consider the predicted motion vectors of the four spatial candidates in the list in a specified order. For example, the candidate's predicted motion vector position 252A may be considered first. If the candidate predicted motion vector position 252A is available, the candidate predicted motion vector position 252A may be assigned to index 0. If the candidate predicted motion vector position 252A is not available, the video decoder may not add the candidate predicted motion vector position 252A to the list of candidate predicted motion vectors. The candidate's predicted motion vector position may not be available for a variety of reasons. For example, if the candidate's predicted motion vector position is not in the current picture, the candidate's predicted motion vector position may not be available. In another feasible implementation, if the candidate's predicted motion vector position goes through intra-prediction, the candidate's predicted motion vector position may not be available. In another feasible implementation, if the candidate's predicted motion vector position is in a different slice than the slice corresponding to the current CU, the candidate's predicted motion vector position may not be available. be.

After considering the candidate predicted motion vector position 252A, the video decoder may consider the candidate predicted motion vector position 252B. If the candidate's predicted motion vector position 252B is available and different from the candidate's predicted motion vector position 252A, the video decoder will place the candidate's predicted motion vector position 252B as the candidate's predicted motion vector position 252B. It may be added to the list of motion vectors. In this particular context, the terms "same" or "different" mean that a candidate's predicted motion vector position is associated with the same or different motion information. Therefore, if the predicted motion vector positions of the two candidates have the same motion information, then the predicted motion vector positions of the two candidates are considered to be the same, or the predicted motion vector positions of the two candidates are If they have different motion information, the predicted motion vector positions of the two candidates are considered to be different. If the candidate predicted motion vector position 252A is not available, the video decoder may assign the candidate predicted motion vector position 252B to index 0. If the predicted motion vector position 252A of the candidate is available, the video decoder, the predicted motion vector position 252 B of the candidate, it is possible to assign the index 1. If the candidate's predicted motion vector position 252B is unavailable or is the same as the candidate's predicted motion vector position 252A, the video decoder skips the candidate's predicted motion vector position 252B. , Do not add the candidate's predicted motion vector position 252B to the list of candidate's predicted motion vectors.

Similarly, the video decoder takes into account the candidate predicted motion vector position 252C and decides whether to add the candidate predicted motion vector position 252C to the list. If a candidate predicted motion vector position 252C is available and is different from the candidate predicted motion vector positions 252B and 252A, the video decoder will use the candidate predicted motion vector position 252C as the next available. Can be assigned to various indexes. If the candidate's predicted motion vector position 252C is not available or is identical to at least one of the candidate's predicted motion vector positions 252A and 252B, the video decoder will perform the candidate's predicted motion vector. Do not add position 252C to the list of candidate predicted motion vectors. The video decoder then considers the candidate predicted motion vector position 252D. If a candidate predicted motion vector position 252D is available and is different from the candidate predicted motion vector positions 252A, 252B, and 252C, the video decoder sets the candidate predicted motion vector position 252D to the next. Can be assigned to the available indexes of. If the candidate's predicted motion vector position 252D is not available or is identical to at least one of the candidate's predicted motion vector positions 252A, 252B, and 252C, the video decoder predicts the candidate. The motion vector position 252D is not added to the list of candidate predicted motion vectors. In the above implementation, the candidate predicted motion vectors 252A to 252D are considered to determine whether the candidate predicted motion vectors 252A to 252D are added to the list of candidate predicted motion vectors. An example of how to do this is outlined. However, in some implementations, all of the candidate predicted motion vectors 252A to 252D may first be added to the list of candidate predicted motion vectors, and then iterative candidate predicted motion. The vector position is removed from the list of candidate predicted motion vectors.

After the video decoder considers the predicted motion vectors of the first four spatial candidates, the list of predicted motion vectors of the candidates may contain the predicted motion vectors of the four spatial candidates, or It may contain predicted motion vectors of less than four spatial candidates. If the list contains the predicted motion vectors of four spatial candidates (yes at 604), i.e., the video decoder considers the time candidates (606), and each time candidate is one. Corresponds to the predicted motion vector of one candidate. The predicted motion vector of the time candidate may correspond to the motion information of the PU at the same position in the picture different from the current picture. If the predicted motion vector of the time candidate is available and different from the predicted motion vector of the first four spatial candidates, the video decoder assigns the predicted motion vector of the time candidate to index 4. If the predicted motion vector of the time candidate is not available or is identical to one of the predicted motion vectors of the first four spatial candidates, the video decoder will use the predicted motion vector of the time candidate. Is not added to the list of candidate predicted motion vectors. Thus, after the video decoder considers the predicted motion vectors of the time candidates (606), the list of predicted motion vectors of the candidates is the first of the five predicted motion vectors considered in 602. It may contain the predicted motion vectors of the four spatial candidates and the predicted motion vectors of the time candidates considered in 604), or the first predicted motion vectors of the four candidates considered in 602. Predicted motion vectors of the four spatial candidates). If the list of candidate predicted motion vectors contains five candidate predicted motion vectors (yes in 608), i.e., the video decoder completes the construction of the list.

If the list of candidate predicted motion vectors contains four candidate predicted motion vectors (No in 608), the video decoder may consider the fifth spatial candidate's predicted motion vector. You can (610). The predicted motion vector of the fifth spatial candidate may (eg) correspond to the predicted motion vector position 252E of the candidate. If a candidate predicted motion vector corresponding to position 252E is available and is different from the candidate predicted motion vector corresponding to positions 252A, 252B, 252C, and 252D, the video decoder is a fifth. The predicted motion vector of the spatial candidate can be added to the list of predicted motion vectors of the candidate, and the predicted motion vector of the fifth spatial candidate can be assigned to the index 4. Candidate predicted motion vectors corresponding to position 252E are not available, or candidate predicted motion vectors of the candidate predicted motion vectors corresponding to positions 252A, 252B, 252C, and 252D . If the one that is the same, the video decoder, the predicted motion vector candidate corresponding to the position 252 may not add to the list of predicted motion vector candidates. Therefore, after the predicted motion vectors of the fifth spatial candidate are considered (610), the list predicts the predicted motion vectors of the five candidates (the first four spatial candidates considered in 602). It may contain motion vectors, and the predicted motion vectors of the fifth spatial candidate considered in 610, or the predicted motion vectors of four candidates (the first four spatial candidates examined in 602). Predicted motion vector) may be included.

If the list of candidate predicted motion vectors contains 5 candidates (yes in 612), that is, if it contains 5 candidate predicted motion vectors, the video decoder will list the candidate predicted motion vectors. Completes the generation of. If the list of candidate predicted motion vectors contains four candidate predicted motion vectors (No at 612), the video decoder will run until the list contains five candidates (yes at 616), i.e. five. Add the manually generated candidate predicted motion vectors until they contain the candidate predicted motion vectors (614).

If the list contains less than four predicted motion vectors of spatial candidates (No in 604) after the video decoder considers the predicted motion vectors of the first four spatial candidates, then the video decoder is the first. The 5 spatial candidates can be considered (618), i.e. the predicted motion vector of the 5th spatial candidate can be considered. The predicted motion vector of the fifth spatial candidate may (eg) correspond to the predicted motion vector position 252E of the candidate. If the candidate predicted motion vector corresponding to position 252E is available and is different from the candidate predicted motion vector present in the list of candidate predicted motion vectors, the video decoder is in the fifth space. Candidate predicted motion vectors can be added to the list of candidate predicted motion vectors and the fifth spatial candidate predicted motion vector can be assigned to the next available index. If the candidate predicted motion vector corresponding to position 252E is not available or is identical to one of the candidate predicted motion vectors present in the candidate's list of predicted motion vectors, video. The decoder may not add the candidate predicted motion vector corresponding to position 252E to the list of candidate predicted motion vectors. The video decoder can then consider temporal candidates, i.e., temporally predicted motion vectors (620). If the predicted motion vector of the time candidate is available and different from the predicted motion vector of the candidate present in the list of predicted motion vectors of the candidate, the video decoder will display the predicted motion vector of the time candidate. Can be added to the list of candidate predicted motion vectors and the time candidate predicted motion vector can be assigned to the next available index. If the predicted motion vector of the time candidate is unavailable or is identical to one of the predicted motion vectors of the candidate present in the list of predicted motion vectors of the candidate, the video decoder will run the time. Candidate predicted motion vectors may not be added to the list of candidate predicted motion vectors.

After the predicted motion vector (618) of the fifth space candidate and the predicted motion vector (620) of the time candidate are considered, the list of predicted motion vectors of the candidate is the fifth space candidate. If it contains a predicted motion vector (yes at 622), the video decoder completes the generation of a list of candidate predicted motion vectors. If the list of candidate predicted motion vectors contains less than 5 candidate predicted motion vectors (No at 622), the video decoder will continue until the list contains 5 candidate predicted motion vectors. (Yes at 616), add the predicted motion vector of the manually generated candidate (614).

According to the technique of the present application, the predicted motion vectors of additional merge candidates can be manually generated after the predicted motion vectors of the spatial candidates and the predicted motion vectors of the time candidates. As a result, the size of the list of predicted motion vectors of the merge candidates will always be the predicted motion vectors of the specified amount of merge candidates (eg, the predictions of the five candidates in the feasible implementation of FIG. 6 above. Is equal to the motion vector). The predicted motion vectors of the additional merge candidates are, for example, the predicted motion vectors of the combined bidirectional predictive merge candidates (candidate predicted motion vector 1), the scaled bidirectional predictive merge candidate predictions. It may include a motion vector (candidate predicted motion vector 2) and a zero vector merge candidate predicted motion vector (candidate predicted motion vector 3). See FIGS. 7-9 for a specific description of the above three cases of predicted motion vectors for additional merge candidates.

FIG. 7 is a schematic diagram of an example of adding a combined candidate motion vector to a list of predicted motion vectors of merge mode candidates according to an embodiment of the present application. The predicted motion vectors of the combined bi-predictive merge candidates can be generated by combining the predicted motion vectors of the original merge candidates. Specifically, the predicted motion vectors of the two original candidates (having mvL0 and refIdxL0, or mvL1 and refIdxL1) can be used to generate the predicted motion vectors of the bipredictive merge candidates. In FIG. 7, the predicted motion vectors of the two candidates are included in the list of predicted motion vectors of the original merge candidates. The prediction type of the predicted motion vector of one candidate is one-way prediction by using List 0, and the prediction type of the predicted motion vector of the other candidate is one-way prediction by using List 1. be. In this feasible implementation, mvL0_A and ref0 are obtained from Listing 0, and mvL1_B and ref0 are obtained from Listing 1. It is then possible to generate the predicted motion vectors of the bi-predictive merge candidates (having mvL0_A and ref0 in Listing 0 and mvL1_B and ref0 in Listing 1), and the predicted motion vectors of the bi-predictive merge candidates. Is checked for different from the candidate's predicted motion vector present in the candidate's list of predicted motion vectors. If the predicted motion vector of a bi-predicted merge candidate is different from the predicted motion vector of an existing candidate, the video decoder will display the predicted motion vector of the bi-predicted merge candidate as a list of the predicted motion vectors of the candidate. Can be added to.

FIG. 8 is a schematic diagram of an example of adding a scaled candidate motion vector to a list of predicted motion vectors of merge mode candidates according to an embodiment of the present application. The predicted motion vector of the scaled bi-predictive merge candidate may be generated by scaling the predicted motion vector of the original merge candidate. Specifically, one original candidate predicted motion vector (with mvLX and refIdxLX) can be used to generate a bipredicted merge candidate predicted motion vector. In the feasible implementation of FIG. 8, the predicted motion vectors of the two candidates are included in the list of predicted motion vectors of the original merge candidates. The prediction type of the predicted motion vector of one candidate is one-way prediction by using List 0, and the prediction type of the predicted motion vector of the other candidate is one-way prediction by using List 1. be. In this feasible implementation, mvL0_A and ref0 are obtained from Listing 0, ref0 is copied to Listing 1 and shown as the reference index ref0'. MvL0'_A can then be calculated by scaling mvL0_A with ref0 and ref0'. Scaling may depend on the POC (Picture Order Count, POC) distance. It is then possible to generate predicted motion vectors of bi-predictive merge candidates (having mvL0_A and ref0 in Listing 0, mvL0'_A and ref0' in List 1) and predicted bi-predictive merge candidates. It is checked whether the motion vector is repeated. If the predicted motion vector of the bi-predicted merge candidate is not repeated, it is possible that the predicted motion vector of the bi-predicted merge candidate is added to the list of predicted motion vectors of the merge candidate.

FIG. 9 is a schematic diagram of an example of adding a zero motion vector to a list of predicted motion vectors of merge mode candidates according to an embodiment of the present application. The predicted motion vector of the zero vector merge candidate can be generated by combining the zero vector and the reference index that can be referenced. If the predicted motion vector of the zero vector candidate is not repeated, the predicted motion vector of the zero vector candidate can be added to the list of predicted motion vectors of the merge candidate. The motion information of the predicted motion vector of each generated merge candidate can be compared with the motion information of the predicted motion vector of the previous candidate in the list.

In a feasible implementation, if the predicted motion vector of the newly generated candidate is different from the predicted motion vector of the candidate present in the list of predicted motion vectors of the candidate, the predicted motion vector of the generated candidate is predicted. The motion vector is added to the list of predicted motion vectors for merge candidates. The process of determining whether a candidate's predicted motion vector differs from the candidate's predicted motion vector present in the candidate's list of predicted motion vectors is often referred to as pruning. Pruning allows each newly generated predicted motion vector of a candidate to be compared with the predicted motion vector of an existing candidate in the list. In some feasible implementations, the pruning process compares the predicted motion vector of one or more new candidates with the predicted motion vector of the candidate present in the list of predicted motion vectors of the candidate. It is possible to include skipping the addition of new candidate predicted motion vectors that are identical to the candidate's predicted motion vectors present in the candidate's list of predicted motion vectors. In some other feasible implementation, the pruning process adds one or more new candidate predicted motion vectors to the list of candidate predicted motion vectors and then iterative candidate predictions. May include excluding motion vectors from the list.

In a feasible implementation of the present application, a method of predicting motion information of a picture block to be processed during interprediction is: at least one picture in which the motion vector is determined in the picture in which the picture block to be processed is located. At least one picture block in which the motion vector is determined in the step of acquiring the motion information of the block is a picture block not adjacent to the picture block to be processed, and the motion vector is determined. A step including a picture block and a step of acquiring the first identification information, wherein the first identification information determines the target motion information in the motion information of at least one picture block in which the motion vector is determined. Includes a step used in the process and a step of predicting the motion information of the picture block to be processed based on the target motion information.

FIG. 10 is a flowchart of an example of the merge mode according to the embodiment of the present application. The video encoder (eg, the video encoder 20) can perform the merge operation 200. In another feasible implementation, the video encoder may perform a different merge operation than the merge operation 200. For example, in another feasible implementation, the video encoder can perform the merge operation, and the video encoder is different from more or less steps in the merge operation 200, or steps in the merge operation 200. Perform the steps. In another feasible implementation, the video encoder can perform the steps of the merge operation 200 in different orders or in parallel. The encoder may further perform merge operation 200 on the PU encoded in skip mode.

After the video encoder initiates the merge operation 200, the video encoder can generate a list of candidate predicted motion vectors for the current PU (202). The video encoder can generate a list of candidate predicted motion vectors for the current PU in a variety of ways. For example, a video encoder can generate a list of candidate predicted motion vectors for the current PU according to one of the technical examples described below in connection with FIGS. 7-9.

As mentioned above, the list of predicted motion vectors for the current PU candidates may include the predicted motion vectors for the time candidates. The predicted motion vector of the time candidate can indicate the motion information of the PUs at the same position in time. PUs at the same position may be spatially located in the picture frame at the same position as the current PU, but may be located in the reference picture instead of the current picture. In the present application, a reference picture including a corresponding temporal PU may be referred to as a related reference picture. In this application, the reference picture index of the associated reference picture may be referred to as the associated reference picture index. As mentioned above, the current picture may be associated with one or more reference picture lists (eg, List 0 and List 1). The reference picture index can indicate a reference picture by indicating the position of the reference picture in the reference picture list. In some feasible implementations, the current picture may be associated with a combined reference picture list.

For some video encoders, the associated reference picture index is the PU's reference picture index that covers the reference index source position associated with the current PU. In these video encoders, the reference index source located relative to the current PU is adjacent to the left of the current PU or above the current PU. In the present application, if the picture block associated with the PU contains a particular location, the PU may "cover" the particular location. For these video encoders, the video encoder may use the reference picture index 0 if the source position of the reference index is not available.

In some examples, the reference index source location associated with the current PU is within the current CU. In these examples, a PU that covers the reference index source position associated with the current PU may be considered available if the PU is above or to the left of the current CU. However, the video encoder may need to access motion information for another PU in the current CU to determine a reference picture that includes a PU in the same position. Thus, these video encoders can use motion information (eg, reference picture index) for PUs belonging to the current CU to generate predicted motion vectors for time candidates for the current PU. In other words, these video encoders can use the motion information of PUs belonging to the current CU to generate predicted motion vectors of temporal candidates. Therefore, the video encoder may not be able to generate a list of candidate predicted motion vectors for the current PU and the PU covering the reference index source position associated with the current PU in parallel. No.

According to the technique of the present application, the video encoder can explicitly set the associated reference picture index without reference to the reference picture index of any other PU. In this way, the video encoder can generate a list of candidate predicted motion vectors in parallel for the current PU and another PU of the current CU. Since the video encoder explicitly sets the associated reference picture index, the associated reference picture index is not based on the motion information of any other PU in the current CU. In some feasible implementations where the video encoder explicitly sets the associated reference picture index, the video encoder sets the associated reference picture index to a fixed, pre-set reference picture. -It can always be set to an index (for example, 0). In this way, the video encoder generates a predicted motion vector of temporal candidates based on the motion information of the co-located PUs in the reference frame indicated by the preset reference picture index. It is possible to add the predicted motion vectors of the temporal candidates to the list of predicted motion vectors of the current CU candidates. The list of candidate predicted motion vectors is also a reference frame list.

In a feasible implementation where the video encoder explicitly sets the associated reference picture index, the video encoder has a syntax structure (eg, picture header, slice header, APS, or other syntax). In the structure), the associated reference picture index can be explicitly signaled. In this feasible implementation, the video encoder signals the decoder with a relevant reference picture index for each LCU (ie, CTU), CU, PU, TU, or another type of subblock. Can be done. For example, the video encoder can signal that the associated reference picture index for each PU of the CU is equal to "1".

In some feasible implementations, the associated reference picture index may be set implicitly rather than explicitly. In these feasible implementations, the video encoder is indicated by a PU reference picture index that covers the position outside the current CU, even if these positions are not exactly adjacent to the current PU. By using the motion information for the PU in the reference picture, it is possible to generate a predicted motion vector for each temporal candidate in the list of predicted motion vectors for the candidate for the PU of the current CU.

After generating a list of candidate predicted motion vectors for the current PU, the video encoder generates a predictive picture block associated with the candidate's predicted motion vector in the list of candidate predicted motion vectors. Can be (204). The video encoder determines the current PU motion information based on the motion information of the indicated candidate's predicted motion vector, and then generates a predictive picture block associated with the candidate's predicted motion vector. Therefore, it is possible to generate a predictive picture block based on one or more reference blocks indicated by the motion information of the current PU. The video encoder can then select one candidate predicted motion vector from the list of candidate predicted motion vectors (206). The video encoder can select candidate predicted motion vectors in various ways. For example, the video encoder can select one candidate predicted motion vector by analyzing the rate distortion cost of each predicted picture block associated with the candidate predicted motion vector.

After selecting a candidate's predicted motion vector, the video encoder can output the candidate's predicted motion vector index (208). The candidate's predicted motion vector index is also the reference frame list index, and the candidate's predicted motion vector index is the candidate's predicted motion vector index selected from the candidate's list of predicted motion vectors. The position of the motion vector can be shown. In some feasible implementations, the candidate predicted motion vector index may be expressed as "merge_idx".

FIG. 11 is a flowchart of an example of the advanced motion vector prediction mode according to the embodiment of the present application. The video encoder (eg, video encoder 20) can perform AMVP operation 210.

After the video encoder initiates the AMVP operation 210, the video encoder can generate one or more motion vectors for the current PU (211). The video encoder can perform integer motion estimation and fractional motion estimation to generate motion vectors for the current PU. As mentioned above, the current picture may be associated with two reference picture lists (List 0 and List 1). If one-way prediction is performed on the current PU, the video encoder can generate a List-0 motion vector or a List-1 motion vector for the current PU. The list-0 motion vector can indicate the spatial displacement between the picture block for the current PU and the reference block of the reference picture in list 0. The Listing-1 motion vector can indicate the spatial displacement between the picture block for the current PU and the reference block of the reference picture in Listing 1. If bidirectional prediction is performed on the current PU, the video encoder can generate a List-0 motion vector and a List-1 motion vector for the current PU.

After generating one or more motion vectors for the current PU, the video encoder may generate a predictive picture block for the current PU (212). The video encoder can generate a predictive picture block for the current PU based on one or more reference blocks indicated by one or more motion vectors for the current PU.

In addition, the video encoder can generate a list of candidate predicted motion vectors for the current PU (213). The video encoder can generate a list of candidate predicted motion vectors for the current PU in a variety of ways. For example, a video encoder may generate a list of candidate predicted motion vectors for the current PU according to one or more feasible implementations described below in connection with FIGS. 6-9. Can be done. In some feasible implementations, if the video encoder produces a list of candidate predicted motion vectors in AMVP operation 210, the list of candidate predicted motion vectors will be two candidate predicted motion vectors. May be limited to motion vectors. In contrast, if the video encoder produces a list of candidate predicted motion vectors in a merge operation, the list of candidate predicted motion vectors will be a list of more candidate predicted motion vectors (eg,). It may contain 5 candidate predicted motion vectors).

After generating a list of candidate predicted motion vectors for the current PU, the video encoder will in the list of candidate predicted motion vectors one or more for each candidate's predicted motion vector. A motion vector difference (MVD) can be generated (214). The video encoder determines the difference between the motion vector indicated by the candidate's predicted motion vector and the corresponding motion vector for the current PU to generate the motion vector difference for the candidate's predicted motion vector. can do.

If one-way prediction is performed on the current PU, the video encoder can generate a single MVD for each candidate predicted motion vector. If bidirectional prediction is performed on the current PU, the video encoder can generate two MVDs for each candidate predicted motion vector. The first MVD can show the difference between the motion vector represented by the candidate predicted motion vector and the list-0 motion vector for the current PU. The second MVD can show the difference between the motion vector represented by the candidate's predicted motion vector and the List-1 motion vector for the current PU.

The video encoder can select one or more candidate predicted motion vectors from the list of candidate predicted motion vectors (215). The video encoder can select one or more candidate predicted motion vectors in various ways. For example, the video encoder can select a candidate predicted motion vector that matches the associated motion vector of the motion vector to be encoded with minimal error. This can reduce the amount of bits required to represent the motion vector difference for the candidate's predicted motion vector.

After selecting one or more candidate predicted motion vectors, the video encoder will have one or more reference picture indexes for the current PU and one or more candidate predicted motion vectors for the current PU. It is possible to output the index and one or more motion vector differences with respect to the predicted motion vector of one or more selected candidates (216).

In an example where the current picture is associated with two reference picture lists (List 0 and List 1) and one-way prediction is performed for the current PU, the video encoder has a reference picture index to List 0 (List 0). A reference picture index (“ref_idx_11”) for “ref_idx_10”) or Listing 1 can be output. The video encoder also provides a candidate predicted motion vector that indicates the position of the selected candidate's predicted motion vector for the list-0 motion vector with respect to the current PU within the list of candidate predicted motion vectors. -The index ("mbp_10_flag") can be output. Alternatively, the video encoder is a candidate's predicted motion vector that indicates the position of the selected candidate's predicted motion vector for the list-1 motion vector with respect to the current PU within the list of candidate's predicted motion vectors. A motion vector index ("mvp_11_flag") can be output. The video encoder can also output a List-0 motion vector for the current PU or an MVD for the List-1 motion vector.

In an example where the current picture is associated with two reference picture lists (List 0 and List 1) and bidirectional prediction is performed for the current PU, the video encoder is for List 0. A reference picture index (“ref_idx_11”) can be output for the reference picture index (“ref_idx_10”) and list 1. In addition, the video encoder indicates the position of the selected candidate's predicted motion vector for the list-0 motion vector with respect to the current PU in the list of candidate's predicted motion vectors. The index ("mvp_10_flag") can be output. In addition, the video encoder indicates the position of the selected candidate's predicted motion vector of the list-1 motion vector with respect to the current PU in the list of candidate's predicted motion vectors. The index ("mvp_11_flag") can be output. The video encoder can also output an MVD for the List-0 motion vector for the current PU and an MVD for the List-1 motion vector for the current PU.

FIG. 12 is a flowchart of an example of motion compensation performed by a video decoder (eg, video decoder 30) according to an embodiment of the present application.

If the video decoder performs a motion compensation operation 220, the video decoder can be instructed by the predicted motion vector of the selected candidate for the current PU (222). For example, the video decoder receives a candidate predicted motion vector index indicating the position of the selected candidate's predicted motion vector in the list of candidate predicted motion vectors for the current PU. be able to.

If the motion information of the current PU is encoded in AMVP mode and bidirectional prediction is performed for the current PU, the video decoder will have the predicted motion vector index of the first candidate and the second candidate. It is possible to receive the predicted motion vector index. The predicted motion vector index of the first candidate indicates the position of the predicted motion vector of the selected candidate of the list-0 motion vector with respect to the current PU in the list of predicted motion vectors of the candidate. The predicted motion vector index of the second candidate indicates the position of the predicted motion vector of the selected candidate of the list-1 motion vector with respect to the current PU in the list of predicted motion vectors of the candidate. In some feasible implementations, a single syntax element can be used to identify the predicted motion vector indexes of the two candidates.

In addition, the video decoder can generate a list of candidate predicted motion vectors for the current PU (224). The video decoder can generate a list of candidate predicted motion vectors for the current PU in a variety of ways. For example, a video decoder can generate a list of candidate predicted motion vectors for the current PU by using the techniques described below in connection with FIGS. 5-9. If the video decoder produces a candidate predicted motion vector in time for a list of candidate predicted motion vectors, the video decoder will be co-located PU as described above with respect to FIG. A reference picture index that identifies the reference picture containing the can be set explicitly or implicitly.

After generating a list of candidate predicted motion vectors for the current PU, the video decoder is indicated by one or more candidate predicted motion vectors in the list of candidate predicted motion vectors for the current PU. Based on the motion information, the motion information for the current PU can be determined (225). For example, if the current PU motion information is encoded in merge mode, the current PU motion information may be the same as the motion information indicated by the predicted motion vector of the selected candidate. If the current PU motion information is encoded in AMVP mode, the video decoder will have one or more motion vectors represented by the predicted motion vectors of the selected candidates and one or more represented by the bitstream. By using with MVD, one or more motion vectors for the current PU can be reconstructed. The current PU reference picture index and predictive direction identifier are identical to one or more reference picture indexes and one or more predictive direction identifiers of one or more selected candidate predicted motion vectors. You may. After determining the motion information for the current PU, the video decoder determines the predicted picture block for the current PU based on one or more reference blocks indicated by the motion information for the current PU. Can be generated (226).

FIG. 13 is an example flowchart for updating the motion vector during video coding according to the embodiment of the present application. The picture block to be processed is a block to be encoded.

1301: Acquires the initial motion vector of the picture block to be processed based on the predicted motion vector of the picture block to be processed.

In a viable implementation, for example, in merge mode, the predicted motion vector of the picture block to be processed is used as the initial motion vector of the picture block to be processed.

The predicted motion vector of the picture block to be processed is the method shown in FIGS. 6 to 9 in the embodiment of the present application, or H. It can be acquired by any existing predicted motion vector acquisition method in the 265 standard or JEM reference mode. This is not limited. The motion vector difference can be acquired in relation to the picture block to be processed. Motion estimation is performed within a search range determined based on the predicted motion vector of the picture block to be processed, and the motion vector of the picture block to be processed and the picture to be processed obtained after motion estimation. The difference between the block and the predicted motion vector is used as the motion vector difference.

During bidirectional prediction, this step specifically: Get the first initial motion vector of the picture block to be processed based on the forward predicted motion vector of the picture block to be processed. And the step of acquiring a second initial motion vector of the picture block to be processed based on the backward predicted motion vector of the picture block to be processed.

1302: Obtains a predicted block of picture blocks to be processed based on the initial motion vector and one or more preset motion vector offsets. Details are as follows:

S13021: The picture block represented by the initial motion vector of the picture block to be processed is acquired from the reference frame of the picture block to be processed indicated by the reference frame index of the picture block to be processed, and the processing target is obtained. It is supplied as a time prediction block of the picture block of.

The initial motion vector includes a first initial motion vector and a second initial motion vector. The first initial motion vector indicates the first motion compensation block based on the first reference frame list of the picture block to be processed, and the first motion compensation block is motion compensation in the picture block to be processed. Is a reference block for executing. The second initial motion vector indicates a second motion compensation block based on the second reference frame list of the picture block to be processed, and the second motion compensation block is motion compensation in the picture block to be processed. Is a reference block for executing.

13022: To get one or more final motion vectors, add the initial motion vector and one or more preset motion vector offsets of the picture block to be processed. Here, each final motion vector indicates a search position.

The preset motion vector offset may be a preset offset vector value, a preset offset vector accuracy, or a preset offset vector range. May be. The specific value of the preset motion vector offset and the amount of the preset motion vector offset are not limited in this embodiment of the invention.

The final motion vector includes a first final motion vector and a second final motion vector. The first final motion vector indicates a first motion compensation block based on the first reference frame list of the picture block to be processed, and the first motion compensation block is motion compensation in the picture block to be processed. Is a predictive block for executing. The second final motion vector indicates a second motion compensation block based on the second reference frame list of the picture block to be processed, and the second motion compensation block is motion compensation in the picture block to be processed. Is a predictive block for executing.

13023: Acquire one or more candidate prediction blocks at one or more search positions indicated by one or more final motion vectors. Here, each search position corresponds to one candidate prediction block.

13024: From one or more candidate prediction blocks, the candidate prediction block with the smallest pixel difference from the temporal prediction block is selected as the prediction block for the picture block to be processed.

The minimum pixel difference may also be understood as the minimum distortion cost. In this case, the final motion vector of the prediction block is the candidate final motion vector corresponding to the minimum strain cost in the final motion vectors of the plurality of candidates. It should be understood that strain costs may be calculated in multiple ways. For example, the sum of the absolute differences between the pixel matrices of the candidate prediction block and the temporal prediction block may be calculated, or the sum of the absolute differences after the pixel matrix averages have been removed may be calculated. Alternatively, another relatively accurate value of the pixel matrix may be calculated. The specific content of the minimum strain cost is not limited to this embodiment of the present invention.

During bidirectional prediction, this step specifically: From the first reference frame list of the picture block to be processed, indicated by the first reference frame list index of the picture block to be processed. In the step of acquiring the first picture block indicated by the first initial motion vector of the picture block of, the first picture block is the forward reference block and correspondingly the first initial motion vector. From the step and the second reference frame list of the picture block to be processed, which may be the initial motion vector ahead, from the second reference frame list of the picture block to be processed. The step of acquiring the second picture block indicated by the second initial motion vector of the picture block, the second picture block being the back reference block, correspondingly the second initial motion vector. Performs weighting on the first and second picture blocks to obtain the steps and the time-predicted block blocks of the picture blocks to be processed, which may be the initial motion vector backwards. A step and a first initial motion vector of a picture block to be processed and one or more preset motion vector offsets to obtain one or more first final motion vectors. To add up to get one or more second final motion vectors, the second initial motion vector of the picture block to be processed and one or more preset motion vector offsets. Obtain one or more first candidate prediction blocks at one or more search positions indicated by the steps to be added and one or more second final motion vectors, and one or more second finals. Candidate prediction with the minimum pixel difference from the time prediction block from the step of acquiring one or more first candidate prediction blocks at one or more search positions indicated by the motion vector and one or more first candidate prediction blocks. A block is selected as the first prediction block of the picture block to be processed, and a candidate prediction block having the smallest pixel difference from the time prediction block from one or more backward candidate prediction blocks is selected as the picture block to be processed. To get the step to select as the second prediction block of, and the prediction block of the picture block to be processed, the first prediction It includes a step of executing a weighting process for the measurement block and the second prediction block.

In some embodiments, another motion vector in the direction indicated by the candidate's final motion vector corresponding to the minimum strain cost in the final motion vector of the plurality of candidates alternatives to the final motion of the predicted block. It may be used as a vector. Specifically, the final motion vector of the prediction block may be selected based on the first preset condition.

In a possible implementation, the final motion vector of the plurality of candidates may be a plurality of motion vectors in multiple directions. First, a particular accuracy in each direction may be selected. For example, the final motion vector of the candidate with integer pixel accuracy is used as the first vector in each direction, and the first vector corresponding to the minimum strain cost is selected as the reference vector from the plurality of first vectors. The first pre-set between the strain cost A corresponding to the reference vector, the strain cost C corresponding to the second vector corresponding to the negative reference vector, and the strain cost B corresponding to the initial motion vector. If the relationship satisfies the first preset condition, the reference vector is used as the final motion vector of the prediction block. The first pre-set between the strain cost A corresponding to the reference vector, the strain cost C corresponding to the second vector corresponding to the negative reference vector, and the strain cost B corresponding to the initial motion vector. If the relationship does not satisfy the first preset condition and A and B satisfy the second preset condition, then the length of the first vector is to obtain the third vector. Decreased with a preset precision, the third vector is used as the first target vector of the prediction block. If A and B do not meet the second preset condition, and B and C satisfy the third preset condition, then the length of the first vector is to obtain the fourth vector. Increased with preset precision, the fourth vector is used as the first target vector of the prediction block. If B and C do not meet the third preset condition, a particular loop count is set. In each loop, the offset of the reference vector is updated first. Specifically, if the second preset relationship between A, B, C satisfies the fourth preset condition, the offset of the reference vector is updated. If the second pre-set relationship between A, B, C does not meet the fourth pre-set condition, or the loop count drops to zero, the offset of the reference vector is not updated. The loop ends. After the loop ends, the direction of the offset is determined based on the relationship between the values of A and C, and the direction of the offset includes the positive or negative direction. A fifth vector is obtained based on the reference vector, the offset, and the direction of the offset, and the fifth vector is used as the first target vector of the prediction block. Finally, the first target vector is used as the final motion vector of the prediction block. The first pre-set relationship may be a proportional relationship, a differential relationship, or a value relationship. The first pre-set condition may be that the first pre-set relationship between A, C, and B is equal to the pre-set value, or the second pre-set. The condition may be that A is equal to B, and the third pre-set condition may be that B is equal to C. 1st pre-set relationship, 1st pre-set condition, pre-set value, 2nd pre-set condition, 3rd pre-set condition, 2nd pre-set relationship , And the fourth pre-set conditions are not particularly limited in this embodiment of the invention.

For example, there are eight candidate motion vectors, and the eight candidate motion vectors are located in four directions: up, down, left, and right of the prediction block. There are two candidate motion vectors in each direction. To obtain the four first vectors, the maximum length candidate motion vector is first selected from each of the four directions. If the first vector in the upward direction corresponds to the minimum strain cost, the first vector in the upward direction is used as the reference vector, that is, the final vector of the prediction block is the positive motion vector. The first pre-set condition is that the proportional relationship between the strain cost A corresponding to the reference vector, the strain cost C corresponding to the second vector, and the strain cost B corresponding to the initial motion vector is 1. Yes, the second vector is a candidate motion vector in the opposite direction to the reference vector, and it is assumed that the preset precision is 1/4 precision. In this case, when the strain cost corresponding to the reference vector is 2, the strain cost corresponding to the first downward vector is 4, and the strain cost corresponding to the initial motion vector is 3, the reference vector. The proportional relationship between the strain cost corresponding to, the strain cost corresponding to the first downward vector, and the strain cost corresponding to the initial motion vector is (2 × B) / (A + C), and the proportional relationship is 1. Is assumed. In this case, the reference vector is used as the final motion vector of the prediction block. If the strain cost corresponding to the first vector in the downward direction is 5 and the ratio relationship (2 × B) / (A + C) is not 1, the ratio does not satisfy the first preset condition. In this case, if A is equal to B, the length of the reference vector is reduced by a preset precision of 1/4 to obtain the third vector, which is the final motion of the prediction block. Used as a vector. If A is not equal to B and B is equal to C, then the length of the reference vector is increased with a preset precision of 1/4 to obtain the 4th vector, which is the prediction block. Used as the final motion vector of. If B is not equal to C, the loop count is set to 3 and the offset accuracy of the reference vector is first updated to 1/8 in each loop and a second advance between A, B, and C. It is assumed that the set relationships are K = | (AC) × 16 | and T = ((A + C) -2 × B) × 8. If K is greater than or equal to T, the fourth preset condition is satisfied and the offset accuracy of the reference vector is updated to 1/16. It is assumed that K and T are updated to K = KT and T = T / 2 in each loop. When the loop count drops to 0, the loop ends. After the loop ends, the direction of the offset is determined based on the relationship between the values of A and C. If A is greater than or equal to C, the offset direction is positive. If A is less than C, the offset direction is negative. A fifth vector is obtained based on the reference vector, the offset, and the direction of the offset, and the fifth vector is used as the first target vector of the prediction block. Finally, the first target vector is used as the final motion vector of the prediction block.

Indeed, in some embodiments, a third pre-set relationship between the strain cost corresponding to the reference vector and the strain cost corresponding to the third vector in the target direction is set in the fifth pre-set. If the above conditions are not met, the length of the first vector can be increased by the preset accuracy until the increased strain cost corresponding to the second target vector meets the second target condition. If the third preset relationship between the strain cost corresponding to the reference vector and the strain cost corresponding to the third vector in the target direction satisfies the fifth preset condition, then the first vector. The length can be reduced by the preset accuracy until the reduced strain cost corresponding to the second target vector meets the target condition. The target direction is any direction other than the direction of the reference vector. The third pre-set relationship, the fifth pre-set condition, and the target condition are not specifically limited in this embodiment of the invention.

In some embodiments, the fourth preset relationship of the strain costs corresponding to the reference vector, the strain cost corresponding to the third vector in the target direction, and the strain cost corresponding to the intermediate vector is the sixth. If the preset condition of is satisfied, the third target vector is determined by increasing the length of the reference vector by the preset accuracy. The fourth pre-set relationship between the strain cost corresponding to the reference vector, the strain cost corresponding to the third vector in the target direction, and the strain cost corresponding to the intermediate vector is the sixth pre-set condition. If not satisfied, the third target vector is determined by reducing the length of the reference vector with a preset precision. The intermediate vector is a motion vector located between the reference vector and the second vector. The process of increasing the reference vector with preset precision and decreasing the reference vector with preset precision has been described above and will not be described in detail here again. The fourth pre-set relationship and the sixth pre-set condition are not specifically limited in this embodiment of the invention.

FIG. 14 is a flowchart of an example of updating a motion vector during video decoding according to an embodiment of the present application. The picture block to be processed is a block to be decoded.

1401: The initial motion vector of the picture block to be processed is acquired based on the predicted motion vector of the picture block to be processed indicated by the index.

In a viable implementation, for example, in merge mode, the predicted motion vector of the picture block to be processed is used as the initial motion vector of the picture block to be processed.

The predicted motion vector of the picture block to be processed is the method shown in FIGS. 6 to 9 in the embodiment of the present application, or H. It can be acquired by any one of the existing predicted motion vector acquisition methods in the 265 standard or JEM reference mode. This is not limited. The motion vector difference can be obtained by analyzing the bitstream.

During bidirectional prediction, this step specifically: obtains the first initial motion vector of the picture block to be processed, based on the forward predicted motion vector of the picture block to be processed. , A step of acquiring a second initial motion vector of the picture block to be processed, based on the backward predicted motion vector of the picture block to be processed.

1402: Obtains a predicted block of picture blocks to be processed based on the initial motion vector and one or more preset motion vector offsets. The details are as follows.

S14021: The picture to be processed from the reference frame of the picture block to be processed indicated by the reference frame index of the picture block to be processed in order to be supplied as a temporal prediction block of the picture block to be processed. Gets the picture block represented by the block's initial motion vector.

14022: To obtain one or more final motion vectors, add the initial motion vector and one or more pre-set motion vector offsets of the picture block to be processed. Here, each final motion vector indicates a search position.

14023: Acquire one or more candidate prediction blocks at one or more search positions indicated by one or more final motion vectors. Each search position corresponds to one candidate prediction block.

14024: From one or more candidate prediction blocks, the candidate prediction block with the first pixel difference from the temporal prediction block is selected as the prediction block for the picture block to be processed.

During bidirectional prediction, this step specifically: From the first reference frame of the picture block to be processed, indicated by the first reference frame index of the picture block to be processed, to the picture block to be processed. Gets the first picture block indicated by the first initial motion vector of, and processes from the second reference frame of the picture block to be processed, which is indicated by the second reference frame index of the picture block to be processed. The step of getting the second picture block, which is indicated by the second initial motion vector of the picture block of interest,
A step of performing weighting processing on the first picture block and the second picture block, and a first initial motion vector and processing in order to obtain a temporally predicted block of the picture block to be processed. Add one or more pre-set motion vector offsets of the picture block of interest to get one or more first final motion vectors, the second initial motion vector and the object to be processed. A step of adding one or more pre-set motion vector offsets of a picture block to obtain one or more second final motion vectors and one or more first finals. Get one or more forward candidate prediction blocks at one or more search positions indicated by a motion vector, and one or more at one or more search positions indicated by one or more second final motion vectors. From the step of acquiring the backward candidate prediction block of and one or more forward candidate prediction blocks, the candidate prediction block having the minimum pixel difference from the temporal prediction block is selected as the forward prediction block of the picture block to be processed. Then, from one or more backward candidate prediction blocks, a step of selecting a candidate prediction block having the minimum pixel difference from the temporal prediction block as a backward prediction block of the picture block to be processed, and a picture to be processed. In order to obtain the predicted block of the block, a step of executing a weighting process on the forward predicted block and the backward predicted block is included.

The implementation of updating the motion vector will be described in detail below by using some specific embodiments. It should be understood that motion vector updates are consistent in encoders and decoders, as described in the coding method of FIG. 13 and the decoding method of FIG. Therefore, the following embodiments will be described only from the encoder or decoder. The implementation in the decoder remains consistent with that in the encoder when the description from the encoder is provided, and the implementation in the encoder is consistent with that in the decoder when the description from the decoder is provided. It should be understood that it maintains.

As shown in FIG. 15, the current decoding block is the first decoding block, and the predicted motion information of the current decoding block can be obtained. The motion vector predictors forward and backward of the current decoded block are (-10, 4) and (5, 6), respectively, and the picture order count (POC) of the current decoded block. ) POC is assumed to be 4. The POC is used to indicate the display order of the pictures, and the POCs of the reference pictures indicated by the index values of the displayed reference pictures of the pictures are 2 and 6, respectively. In this case, the POC corresponding to the current decryption block is 4, the POC corresponding to the forward predictive reference picture block is 2, and the POC corresponding to the backward predictive reference picture block is 6.

A forward prediction and backward prediction is performed separately for the current decoding block, the initial pre Ho予the current decoding block measuring block (forward prediction block, FPB) and early after Ho予measuring block ( backward prediction block, BPB). It is assumed that the initial forward decoding prediction block and the initial backward decoding prediction block are FPB1 and BPB1, respectively. The first decoding prediction block (DPB) of the current decoding block is obtained by weighting addition to FPB1 and BPB1 and is assumed to be DPB1.

(-10,4) and (5,6) are used as reference inputs for forward and backward motion vector predictors, with first precision motion search for forward and backward predictive reference picture blocks. Are executed separately. In this case, the first precision is 1/2 pixel precision in the 1 pixel range. The first decryption prediction block DPB1 is used as a reference. The corresponding new forward and backward decoding prediction blocks obtained by each motion search are compared with the first decoding prediction block DPB1 to obtain a new decoding prediction block with the smallest difference from DPB1 and a new decoding. The forward and backward motion vector predictors corresponding to the transformation prediction blocks are used as target motion vector predictors and are assumed to be (-11,4) and (6,6), respectively.

The target motion vector predictor is updated to (-11,4) and (6,6), and forward and backward predictions are performed separately for the first decoding block based on the target motion vector predictor. , The target decryption prediction block is acquired by weighting the acquired new forward and backward decoding prediction blocks and is assumed to be DPB2, and the decoding prediction block of the current decoding block is Updated to DPB2.

If the first precision motion search is performed on the forward predictive reference picture block and the backward predictive reference picture block, the first precision may be any specified precision, eg, an integer pixel precision. , 1/2 pixel accuracy, 1/4 pixel accuracy, or 1/8 pixel accuracy.

As shown in FIG. 16, the current decoding block is the first decoding block, and the predicted motion information of the current decoding block is acquired. The forward motion vector predictor of the current decoded block is (-21,18), the POC of the picture in which the current decoded block is located is 4, and the POC of the reference picture indicated by the index value of the reference picture is It is assumed to be 2. In this case, the POC corresponding to the current decryption block is 4 and the POC corresponding to the forward predictive reference picture block is 2.

Forward prediction is performed on the current decryption block to get the initial forward prediction block of the current decryption block. It is assumed that the initial forward decoding prediction block is FPB1. In this case, the FPB1 is used as the first decoding prediction block of the current decoding block, and the first decoding prediction block is shown as DPB1.

(-21, 18) are used as reference inputs for the forward motion vector predictor, and a first-precision motion search is performed on the forward motion prediction reference picture block. In this case, the first precision is 1 pixel precision in the 5 pixel range. The first decryption prediction block DPB1 is used as a reference. The corresponding new forward decoding prediction block acquired by each motion search is compared with the first decoding prediction block DPB1 to obtain a new decoding prediction block with the minimum difference from DPB1 and a new decoding prediction block. The forward motion vector predictor corresponding to is used as the target motion vector predictor and is assumed to be (-19,19).

The target motion vector predictor is updated to (-19, 19), forward prediction is performed based on the target motion vector predictor, and the new forward decoding prediction block obtained is used as the target decoding prediction block. Assumed to be DPB2, the decoding prediction block of the current decoding block is updated to DPB2.

If the first precision motion search is performed on the forward predictive reference picture block and the backward predictive reference picture block, the first precision may be any specified precision, eg, an integer pixel precision. It should be noted that the accuracy may be 1/2 pixel accuracy, 1/4 pixel accuracy, or 1/8 pixel accuracy.

As shown in FIGS. 17A and 17B, the current coding block is the first coding block, and the predicted motion information of the current coding block can be obtained. The forward and backward motion vector predictors of the current coding block are (-6,12) and (8,4), respectively, and the POC of the picture in which the current coding block is located is 8, and the index of the reference picture. It is assumed that the POCs of the reference pictures indicated by the values are 4 and 12, respectively. In this case, the POC corresponding to the current coding block is 4, the POC corresponding to the forward predictive reference picture block is 4, and the POC corresponding to the backward predictive reference picture block is 12.

The forward and backward predictions are performed separately for the current coded block to get the initial forward and backward coded predictive blocks of the current coded block. The initial forward-coded predictive block and the initial backward-coded predictive block are assumed to be FPB1 and BPB1, respectively. The first coding prediction block of the current coding block is obtained by performing weighting addition to FPB1 and BPB1 and is assumed to be DPB1.

(-6,12) and (8,4) are used as reference inputs for forward and backward motion vector predictors, with first precision motion search for forward and backward predictive reference picture blocks. Will be executed separately. The first coded prediction block DPB1 is used as a reference. The corresponding new forward and backward coded predictive blocks obtained by each motion search are compared with the first coded predictive block DPB1 to obtain a new coded predictive block with the minimum difference from the DBI and the new coded. The forward and backward motion vector predictors corresponding to the prediction blocks are used as target motion vector predictors and are assumed to be (-11,4) and (6,6), respectively.

The target motion vector predictors are updated to (-11,4) and (6,6), forward and backward predictions are performed separately for the first coded block based on the target motion vector predictors. The target coded predictive block is obtained by performing a weighted addition on the new forward and backward coded predictive blocks obtained and is assumed to be DPB2, and the coded predictive block of the current coding block is DPB2. Will be updated to.

(-11,4) and (6,6) are then used as reference inputs for the forward and backward motion vector predictors, and the first precision motion search is the forward predictive reference picture block and the backward predictive reference picture. Executed separately for blocks. The coded prediction block DPB2 of the current coding block is used as a reference. The corresponding new forward and backward coded prediction blocks obtained in each motion search are compared with the first coded predictive block DPB2 to obtain a new coded predictive block with the smallest difference from DPB2 and a new code. The forward and backward motion vector predictors corresponding to the transformation prediction blocks are used as new target motion vector predictors and are assumed to be (-7,11) and (6,5), respectively.

The target motion vector predictors are then updated to (-7,11) and (6,5), and forward and backward predictions are based on the latest target motion vector predictors for the first coding block. Executed separately, the target coded predictive block is obtained by performing a weighted addition on the new forward and backward coded predictive blocks obtained, assumed to be DPB3, and the code of the current coding block. The conversion prediction block is updated to DPB3.

Further, the target motion vector predictor can be continuously updated according to the above method, and the number of cycles is not limited.

If the first precision motion search is performed on the forward predictive reference picture block and the backward predictive reference picture block, the first precision may be any specified precision, eg, integer pixel precision, It should be noted that it may be 1/2 pixel accuracy, 1/4 pixel accuracy, or 1/8 pixel accuracy.

It should be understood that in some feasible embodiments, the cycle ends when the conditions are met. For example, the cycle ends when the difference between DPBn and DPBn-1 is less than the threshold, where n is a positive integer greater than 2.

As shown in FIG. 18, the current decoding block is the first decoding block, and the predicted motion information of the current decoding block can be obtained. The forward and backward motion vector values of the current decoded block are (-10,4) and (5,6), respectively, and the forward and backward motion vector differences of the current decoded block are (-2,1), respectively. And (1,1), it is assumed that the POC of the picture in which the current decoding block is located is 4 and the POC of the reference picture indicated by the index value of the reference picture is 2 and 6, respectively. Therefore, the POC corresponding to the current decoded block is 4, the POC corresponding to the forward predictive reference picture block is 2, and the POC corresponding to the backward predictive reference picture block is 6.

Forward and backward predictions are performed separately for the current decoding block to get the initial forward decoding prediction block (FPB) and the initial backward decoding prediction block (BPB) of the current decoding block. do. The initial forward decoding prediction block and the initial backward decoding prediction block are assumed to be FPB1 and BPB1, respectively. The first decoding prediction block (DPB) of the current decoding block is obtained by performing weighting addition on FPB1 and BPB1 and is assumed to be DPB1.

The sum of the forward motion vector predictor and the forward motion vector difference and the sum of the backward motion vector predictor and the backward motion vector difference are (-10,4) + (-2,1) = (-12,5). And (5,6) + (1,1) = (6,7) are used as forward motion vector and backward motion vector, respectively, and the first-precision motion search refers to the forward prediction reference picture block and the backward prediction reference. Executed separately for picture blocks. In this case, the first precision is 1/4 pixel precision in 1 pixel / range. The first decryption prediction block DPB1 is used as a reference. The corresponding new forward and backward decoding prediction blocks obtained in each motion search are compared with the first decoding prediction block DPB1 to obtain a new decoding prediction block with the minimum difference from DPB1. The forward and backward motion vectors corresponding to the new decoded prediction block are used as target motion vector predictors and are assumed to be (-11,4) and (6,6), respectively.

The target motion vector is updated to (-11,4) and (6,6), forward prediction and backward prediction are executed separately for the first decoding block based on the target motion vector, and the target decoding prediction is performed. The block is acquired by performing weighting addition on the acquired new forward and backward decoding prediction blocks, is assumed to be DPB2, and the decoding prediction block of the current decoding block is updated to DPB2. ..

In order to further reflect the core technical content of the present application, the process of storing motion vectors by the encoder in the merge mode will be described below by using specific embodiments. FIG. 19 is a schematic flowchart of a method of acquiring a motion vector by an encoder according to the embodiment of the present application. The method includes the following steps.

1901: After acquiring the first processed picture block, the encoder determines one or more pre-set candidate reference blocks of the first processed picture block.

Referring to FIG. 3, the first processed picture block is a coded picture block, that is, the coded picture block is such that the partition unit of the encoder partition stores the video data. Obtained after acquisition. It should be noted that the picture block to be processed may be the current PU or the current CU. This is not specifically limited in this embodiment of the invention.

The candidate reference block includes a picture block having a spatial positional relationship set in advance with the picture block to be processed. The preset spatial positional relationship may be the positional relationship shown in FIG. Since the candidate reference block shown in FIG. 5 is adjacent to the picture block to be processed, that is, the picture information of the candidate reference block is compared with that of the picture block to be processed. Due to their similarities, the motion vector of the candidate reference block is used as the predicted motion vector of the first processed picture block in order to predict the first processed picture block. It is possible. In some embodiments, the candidate reference block is further a virtual picture block having a pre-set spatial correlation with another real picture block or a first processed picture block. include. In this embodiment of the present application, the positional relationship between the candidate reference block and the first processed picture block is not specifically limited by a particular amount of candidate reference blocks.

1902: The encoder determines one or more motion vectors of one or more candidate reference blocks.

In this embodiment of the present application, the motion vector update process is executed for the candidate reference block. When the update process for the candidate reference block is completed, the candidate reference block has an initial motion vector and a final motion vector, that is, the candidate reference block has a corresponding initial motion vector and final motion. Memorize the vector. In some embodiments, the candidate reference block stores the initial motion vector and motion vector residuals accordingly, and the final motion vector of the candidate reference block is the initial motion vector of the candidate reference block. And can be obtained by adding the motion vector residuals.

However, if the update process for the candidate reference block is not completed, the candidate reference block has only the initial motion vector, that is, the candidate reference block correspondingly stores only the initial motion vector of the candidate reference block. ..

Since the candidate reference block has an initial motion vector and a final motion vector, the encoder uses the initial motion vector or the final motion vector of the candidate reference block as the motion vector of the candidate reference block. As a result, the motion vector of the candidate reference block is used as the predicted motion vector of the spatial candidate of the picture block to be processed.

It should be noted that for the motion vector update process in the reference block, refer to the embodiment of the present application related to FIG. It should be understood that the reference block of the embodiment related to FIG. 19 is the picture block to be processed in the embodiment related to FIG.

In step 1902, the motion vector of the candidate reference block may be determined based on the previously set range of the candidate reference block. For example, if the candidate reference block is within the preset range, the initial motion vector of the candidate reference block is used as the motion vector of the candidate reference block, or the candidate reference block is preset. If it is out of range, the final motion vector of the candidate reference block is used as the motion vector of the candidate reference block. Therefore, one or more motion vectors of one or more candidate reference blocks can be determined.

Specifically, the preset range may be the CTB row range. For example, if the CTB where the candidate reference block is located and the CTB where the first picture block to be processed is located are located on the same row, the initial motion vector of the candidate reference block is the candidate reference block. Used as a motion vector. Correspondingly, if the CTB where the candidate reference block is located and the CTB where the first picture block to be processed is located are not located in the same row, the final motion vector of the candidate reference block will be: Used as a motion vector for candidate reference blocks. Referring to FIG. 21, the coding tree block in which the candidate reference block is located and the coding tree block in which the first picture block to be processed is located are located on different rows, and the candidate reference block is located. If the located coding tree block and the coding tree block in which the first processed picture block is located are located in different rows of adjacent spaces, then the initial motion vector of the candidate reference block is: It will be appreciated that it will be used as a motion vector for the candidate reference block. The adjacent space of different rows may be the upper space, upper left space, etc. of different rows. The specific range of adjacent spaces in different rows is not specifically limited in this embodiment of the invention.

In a related technique, the encoder may encode a plurality of picture blocks to be processed at the same time. However, when a plurality of processed picture blocks are located in the same CTB row, the plurality of processed picture blocks may be reference blocks of each other. If multiple processed picture blocks are reference blocks of each other's candidates, the updated final motion vector of another block will only be the current processed picture block after another block has been updated. It can be used as a predicted motion vector for block spatial candidates. As a result, there is a coding delay. Further, since the plurality of picture blocks to be processed are reference blocks of each other's candidates, it is necessary to wait until the other blocks are updated. As a result, the coding delay increases.

Therefore, the initial motion vector of the candidate reference block can be used directly as the motion vector of the candidate reference block, and then the motion vector of the candidate reference block is the first object to be processed. Used as a predicted motion vector for spatial candidates in a picture block. See step 1903 for this.

If the CTB where the candidate reference block is located and the CTB where the first picture block to be processed is located are located on the same row, the update process for the candidate reference block may not be completed. Therefore, the initial motion vector of the candidate reference block can be used directly as the motion vector of the candidate reference block, and there is no need to wait for the candidate reference block to be updated, and it is coded. Reduce delay. If the CTB where the candidate reference block is located and the CTB where the first process target picture block is located are not located on the same row, it is sufficient to complete the update process for the candidate reference block. Show that you have time. Therefore, the candidate reference block has a final motion vector, and the final motion vector of the candidate reference block can be used as the motion vector of the candidate reference block. Furthermore, since the final motion vector is the motion vector obtained after the initial motion vector has been refined, using the final motion vector as the motion vector of the candidate reference block guarantees the coding quality. can do.

The preset range may be the CTB block range instead. For example, when the candidate reference block and the first processed picture block are located in the same CTB, the initial motion vector of the candidate reference block is used as the motion vector of the candidate reference block. Correspondingly, if the candidate reference block and the first processed picture block are not located in the same CTB, the final motion vector of the candidate reference block is the motion vector of the candidate reference block. Used as.

If the candidate reference block and the first processed picture block are located in the same CTB, it indicates that the update process in the candidate reference block may not be completed, and the candidate reference block. The initial motion vector of is used directly as the motion vector of the candidate reference block without having to wait for the candidate reference block to be updated, reducing the coding delay. Further, if the candidate reference block and the first processed picture block are not located in the same CTB, it indicates that there is sufficient time to complete the update process for the candidate reference block. .. Further, since the final motion vector is a motion vector obtained after the initial motion vector is refined, the final of the candidate reference blocks that are not located in the same CTB as the first processed picture block. Using the motion vector as the motion vector of the candidate reference block can guarantee the coding quality.

The pre-set range may optionally be the same CTB block range or the range of left-adjacent and right-adjacent CTB blocks. For example, the CTB where the candidate reference block is located is the CTB where the first process target picture block is located, or the CTB where the candidate reference block is located is the first process target picture block. If it is a left-adjacent or right-adjacent block of the located CTB, the initial motion vector of the candidate reference block is used as the motion vector of the candidate reference block. Correspondingly, the CTB where the candidate reference block is located is not the same as the CTB where the first process target picture block is located, or the CTB where the candidate reference block is located is the first process target picture. If the block is not on the left or right side of the CTB where the block is located, the final motion vector of the candidate reference block is used as the motion vector of the candidate reference block.

If the CTB where the candidate reference block is located is the same as the CTB where the first process target picture block is located, or the CTB where the candidate reference block is located is where the first process target picture block is located. If it is a left-adjacent or right-adjacent block of the candidate CTB, it indicates that the update process for the candidate reference block may not be completed, and the initial motion vector of the candidate reference block is updated by the candidate reference block. It is used directly as the motion vector of the candidate reference block without the need to wait until it is done, reducing the coding delay. Further, the CTB where the candidate reference block is located is not the same as the CTB where the first process target picture block is located, or the CTB where the candidate reference block is located is the first process target picture block. If is not a left-adjacent or right-adjacent block of the CTB in which it is located, it indicates that there is sufficient time to complete the update process for the candidate reference block. Furthermore, since the final motion vector is the motion vector obtained after the initial motion vector has been refined, it is not possible to use the final motion vector of the candidate reference block as the motion vector of the candidate reference block. , The coding quality can be guaranteed.

1903: The encoder sets a list of predicted motion vectors of candidates based on one or more motion vectors of one or more candidate reference blocks.

See FIGS. 6-9 for how to set a list of predicted motion vectors for candidates. Details will not be explained here again.

1904: The encoder determines the reference block of the picture block to be processed. The reference block and the first processed picture block are located within the frame of the same picture.

The candidate reference block with the minimum rate distortion cost is selected from one or more candidate reference blocks of the first processable picture block as the reference block of the first processable picture block.

1905: The encoder stores the motion vector of the reference block. Motion vector of the reference block is the initial motion vectors of picture blocks in the first processing target.

The motion vector of the reference block can be obtained from the corresponding position in the list of predicted motion vectors of the set candidates.

Steps 1901 to 1905 are processes in which the encoder acquires the initial motion vector of the first process target picture block.

1906: The encoder determines the predictive block of the first processed picture block based on the initial motion vector of the first processed picture block and one or more preset motion vector offsets. decide.

Step 1906 can be completed by using 13021 to 13024 of FIG. Details will not be explained here again.

One or more pre-set motion vector offsets can be set on a pixel-by-pixel basis. For example, the picture block represented by the initial motion vector of the picture block to be processed is used as a temporal prediction block of the picture block to be processed, and is one of the temporal prediction blocks. One or more candidate prediction blocks within pixel units or 1/2 pixel units are searched by using the temporal prediction blocks as the center. Indeed, different lengths or different pixel units may be used. One or more pre-set motion vector offsets are not specifically limited in this embodiment of the present application.

1907: After determining the motion vector of the predicted block, the encoder uses the motion vector of the predicted block as the final motion vector of the picture block to be processed.

In step 1907, the motion vector of the prediction block may be determined based on the advance the set range before the prediction block. For example, if the predicted block is within a preset range, the initial motion vector of the predicted block is used as the motion vector of the predicted block, and the predicted block is outside the preset range. The final motion vector of the predicted block is used as the motion vector of the predicted block.

Specifically, the preset range may be the CTB row range. For example, if the CTB where the prediction block is located and the CTB where the first picture block to be processed is located are located on the same row, the initial motion vector of the prediction block is used as the motion vector of the prediction block. Will be done. Correspondingly, if the CTB where the prediction block is located and the CTB where the first picture block to be processed is located are not located in the same row, the initial motion vector of the prediction block is the motion vector of the prediction block. Used as. Referring to FIG. 21, the coding tree block in which the prediction block is located and the coding tree block in which the first processing target picture block is located are located on different rows, and the prediction block is located in the coding tree block. If the tree block and the coding tree block in which the first processed picture block is located are located in adjacent spaces of different rows, the initial motion vector of the predicted block is the motion of the predicted block. It will be understood that it is used as a vector. The adjacent space of different rows may be the upper space, upper left space, etc. of different rows. The particular range of adjacent spaces in different rows is not specifically limited in this embodiment of the invention.

If the CTB where the prediction block is located and the CTB where the picture block to be processed first is located are located on the same row, the update processing for the prediction block may not be completed. Therefore, the initial motion vector of the predicted block can be used directly as the motion vector of the predicted block, and it is not necessary to wait until the predicted block is updated, which reduces the coding delay. If the CTB where the prediction block is located and the CTB where the first processing target picture block is located are not located on the same row, it has enough time to complete the update process for the prediction block. Show that. Therefore, the prediction block has a final motion vector, and the final motion vector of the prediction block can be used as the motion vector of the prediction block. Furthermore, since the final motion vector is the motion vector obtained after the initial motion vector has been refined, using the final motion vector as the motion vector of the prediction block guarantees the coding quality. Can be done.

In some embodiments, the preset range may be the CTB block range instead. In some other embodiments, the pre-set range may be the same CTB block instead. Determining the motion vector of the predictive block is similar to determining the motion vector of the candidate reference block. See step 1902 for this. Details will not be explained here again.

1908: The encoder updates the initial motion vector of the first processed picture block in the candidate list to the final motion vector of the first processed picture block.

1909: The encoder stores the motion vector residual of the picture block to be processed in the target storage space.

The motion vector residual is the motion vector difference between the final motion vector of the first processed picture block and the initial motion vector of the first processed picture block.

The target storage space is a space for storing the motion vector residuals by the encoder. For example, the target storage space may be the reference picture storage storage 64 of FIG.

It should be noted that in merge mode in related technology, the target storage space stores zero vectors or the target storage space does not store data. However, in the merge mode of this embodiment of the present application, the motion vector residuals of the picture block to be processed are stored in the target storage space, and as a result, the target storage space can be completely used. Not only that, it is also possible to flexibly select the motion vector of the other picture block when the motion vector of the other picture block is required for the first processed picture block. .. As shown in steps 1902 and 1907, the coding efficiency of the encoder is further improved.

Further, in some implementations, in step 1909, the final motion vector of the first processed picture block is instead used as the motion vector residual of the first processed picture block. , That is, the final motion vector of the picture block to be processed is stored in the target storage space, that is, the picture block to be processed first. The motion vector residuals of are not stored in the target storage space. In this case, if the final motion vector of the picture block to be processed is required for another block, the final motion vector of the picture block to be processed is the first. Can be obtained directly from the encoder's target storage space without having to add the initial motion vector of the picture block to be processed and the motion vector residuals of the first picture block to be processed. It is possible.

See FIG. 21 for the specific storage method of step 1909.

It should be noted that there are no strict requirements for the sequences of steps 1908 and 1909, and the encoder may optionally perform step 1909 prior to step 1908. This is not specifically limited in this embodiment of the present application.

Steps 1906 to 1909 are processes in which the encoder acquires the final motion vector of the first process target picture block.

1910: The encoder encodes one or more first identification information corresponding to the initial motion vector of the picture block to be processed or the initial motion vector of the picture block to be processed in the candidate list into a bitstream. To become.

The identification information may include an index. Indeed, in some implementations, the identification information may further include a predictive mode. The index is used to indicate the corresponding motion vector in the list of candidate predicted motion vectors, which is the initial motion vector or final motion vector of the first processed picture block. , Prediction modes include merge mode, AMVPA mode, and skip mode. In this embodiment of the present application, the prediction mode is the merge mode. Indeed, in another embodiment, the prediction mode may be another mode. This is not limited to this embodiment of the present application.

In the above implementation, when encoding the current coding block, the encoder replaces the final motion vector of the reference block with the unupdated initial motion vector of the reference block to the current coding block. Used as the predicted motion vector of. In this way, if the motion vector of the reference block is needed for the current coding block, the relevant steps need to wait until the final motion vector of the reference block is updated. It can be performed without, while compensating for a reduction in coding delay, while updating the motion vector to improve coding efficiency.

When the inter-prediction mode used by the encoder is forward or backward prediction, the motion vector of the candidate reference block, the motion vector of the reference block, and the motion vector of the predictive block shall include the forward or backward motion vector. Should be noted. Correspondingly, when a list of candidate predicted motion vectors is set, the list of candidate predicted motion vectors contains only list 0 or list 1 and is the initial motion vector of the picture block to be processed. Contains an early forward motion vector or an early backward motion vector, and the final motion vector of the picture block to be processed includes a final forward motion vector or a final backward motion vector.

When the inter-prediction mode used by the encoder is bidirectional prediction, the motion vector of the candidate reference block, the motion vector of the reference block, and the motion vector of the prediction block include the first motion vector and the second motion vector. , The first motion vector is a forward motion vector, and the second motion vector is a backward motion vector. Correspondingly, when a list of candidate predicted motion vectors is set, the list of candidate predicted motion vectors includes list 0 and list 1, and the initial motion vector of the picture block to be processed is , A first initial motion vector and a second initial motion vector, and the final motion vector of the picture block to be processed is the first final motion vector and the second final motion vector. including. The first initial motion vector and the first final motion vector are forward motion vectors, and the second initial motion vector and the second final motion vector are backward motion vectors.

The above content describes the encoder. Correspondingly, the process of storing motion vectors by the decoder in the merge mode will be described below using specific embodiments. FIG. 20 is a schematic flowchart of a method of acquiring a motion vector by a decoder according to the embodiment of the present application. The method includes the following steps.

2001: After receiving the bitstream corresponding to the picture block to be processed, the decoder analyzes the bitstream to obtain the second identification information and the third identification information. Here, the second discriminant information is used to determine the initial predicted motion vector of the picture block to be processed, and the third discriminant information is the motion vector residuals analyzed during decoding. Used to indicate if it is necessary.

The second identification information may include an index used to determine the initial motion vector of the picture block to be processed. Referring to steps 2002 and 2004, in some embodiments, the second identification information may further include a predictive mode type.

The third identification information is used to indicate whether the motion vector residuals need to be analyzed during decoding. In this embodiment of the present application, the motion vector residuals do not need to be analyzed in merge mode, but in another embodiment the motion vector residuals may be analyzed. For example, in the SMVP mode, when the motion vector residual of the picture block to be processed in the second process is added to the bitstream according to the instruction of the third identification information, the decoder analyzes the bitstream to obtain the motion vector residual. Can be obtained. In this case, the decoder stores the motion vector residuals directly in the target storage space.

2002: After receiving the bitstream corresponding to the second processed picture block, the decoder has one or more candidate reference blocks and one or more candidate for the second processed picture block. Determine with one or more motion vectors of the reference block.

The reference block of one or more candidates includes a picture block having a spatial positional relationship set in advance with the picture block to be processed. See FIG. 5 for determining candidate reference blocks. Details will not be explained here again.

In this embodiment of the present application, motion vector update processing is performed on the candidate reference block. When the update process for the candidate reference block is completed, the candidate reference block has an initial motion vector and a final motion vector, that is, the candidate reference block has an initial motion vector and a final motion vector. And remember. In some embodiments, the candidate reference block stores the initial motion vector and motion vector residuals, and the final motion vector of the candidate reference block is the initial motion vector and motion of the candidate reference block. It can be obtained by adding a vector residual.

However, if the update process of the candidate reference block is not completed, the candidate reference block has only the initial motion vector, that is, the candidate reference block stores only the initial motion vector of the candidate reference block. do.

Since the candidate reference block has an initial motion vector and a final motion vector, the decoder uses the initial motion vector or the final motion vector of the candidate reference block as the motion vector of the candidate reference block. As a result, the motion vector of the candidate reference block is used as the predicted motion vector of the spatial candidate of the second processed picture block.

It should be noted that for the motion vector update process for the reference block and the acquisition of the initial motion vector, see embodiments of the present application related to FIG. It should be understood that the reference block of the embodiment related to FIG. 20 is the picture block to be processed according to the embodiment of FIG.

In step 2002 , the motion vector of the candidate reference block may be determined based on the previously set range of the candidate reference block. For example, if the candidate reference block is within the preset range, the initial motion vector of the candidate reference block is used as the motion vector of the candidate reference block and the candidate reference block is preset. If it is out of range, the final motion vector of the candidate reference block is used as the motion vector of the candidate reference block. Therefore, one or more motion vectors of one or more candidate reference blocks can be determined.

Specifically, the preset range may be the CTB row range. For example, if the CTB where the candidate reference block is located and the CTB where the second picture block to be processed is located are located on the same row, the initial motion vector of the candidate reference block is the candidate reference block. Used as a motion vector. Correspondingly, if the CTB where the candidate reference block is located and the CTB where the second picture block to be processed is located are not located on the same row, the initial motion vector of the candidate reference block is a candidate. Used as a motion vector for the reference block of. Referring to FIG. 21, the coding tree block in which the candidate reference block is located and the coding tree block in which the second picture block to be processed is located are located on different rows, and the candidate reference block is located. If the coding tree block in which it is located and the coding tree block in which the second picture block to be processed is located are located in adjacent spaces on different rows, the initial motion vector of the candidate reference block is It will be appreciated that it is used as a motion vector for candidate reference blocks. The adjacent space of different rows may be the upper space, upper left space, etc. of different rows. The specific range of spaces adjacent to different rows is not specifically limited in this embodiment of the invention.

In a related technique, the decoder may simultaneously decode a plurality of picture blocks to be processed. However, when a plurality of processed picture blocks are located in the same CTB row, the plurality of processed picture blocks may be reference blocks of each other. If multiple processed picture blocks are reference blocks of each other's candidates, the updated final motion vector of another block will only be the current processed picture block after another block has been updated. It can be used as a predicted motion vector for block spatial candidates. As a result, there is a decryption delay. Further, since the plurality of picture blocks to be processed are reference blocks of each other's candidates, it is necessary to wait until the other blocks are updated. As a result, the decryption delay increases.

Therefore, the initial motion vector of the candidate reference block can be used directly as the motion vector of the candidate reference block, and then the motion vector of the candidate reference block is the second object to be processed. Used as a predicted motion vector for spatial candidates in a picture block. See step 2003 for this.

If the CTB where the candidate reference block is located and the CTB where the second picture block to be processed is located are located on the same row, the update process for the candidate reference block may not be completed. Therefore, the initial motion vector of the candidate reference block can be used directly as the motion vector of the candidate reference block, without having to wait for the candidate reference block to be updated, and decoding. Reduce delay. If the CTB where the candidate reference block is located and the CTB where the second processed picture block is located are not located on the same row, it is sufficient to complete the update process for the candidate reference block. Show that you have time. Therefore, the candidate reference block has a final motion vector, and the final motion vector of the candidate reference block can be used as the motion vector of the candidate reference block. Furthermore, since the final motion vector is the motion vector obtained after the initial motion vector has been refined, using the final motion vector as the motion vector of the candidate reference block guarantees decoding quality. can do.

The preset range may be the CTB block range instead. For example, when the candidate reference block and the second processed picture block are located in the same CTB, the initial motion vector of the candidate reference block is used as the motion vector of the candidate reference block. Correspondingly, if the candidate reference block and the second processed picture block are not located in the same CTB, the final motion vector of the candidate reference block is the motion vector of the candidate reference block. Used as.

If the candidate reference block and the second processed picture block are located in the same CTB, it indicates that the update process in the candidate reference block may not be completed, and the candidate reference block. The initial motion vector of is used directly as the motion vector of the candidate reference block without having to wait for the candidate reference block to be updated, reducing the decoding delay. Further, if the candidate reference block and the second processed picture block are not located in the same CTB, it indicates that there is sufficient time to complete the update process for the candidate reference block. .. Further, since the final motion vector is a motion vector obtained after the initial motion vector is refined, the final of the candidate reference blocks that are not located in the same CTB as the second processed picture block. Using the motion vector as the motion vector of the candidate reference block can guarantee the decoding quality.

The preset range may optionally be the same CTB block range, or the range of left-adjacent and right-adjacent CTB blocks. For example, the CTB where the candidate reference block is located is the CTB where the second process target picture block is located, or the CTB where the candidate reference block is located is the second process target picture block. If it is a left-adjacent or right-adjacent block of the located CTB, the initial motion vector of the candidate reference block is used as the motion vector of the candidate reference block. Correspondingly, the CTB where the candidate reference block is located is not the same as the CTB where the second process target picture block is located, or the CTB where the candidate reference block is located is the second process target picture. If the block is not on the left or right side of the CTB where the block is located, the final motion vector of the candidate reference block is used as the motion vector of the candidate reference block.

If the CTB where the candidate reference block is located is the same as the CTB where the second process target picture block is located, or the CTB where the candidate reference block is located is the second process target picture block. If it is a left-adjacent or right-adjacent block of the located CTB, it indicates that the update process for the candidate reference block may not be completed, and the initial motion vector of the candidate reference block is that the candidate reference block It is used directly as the motion vector of the candidate reference block without having to wait for it to be updated, reducing the decoding delay. Further, the CTB where the candidate reference block is located is not the same as the CTB where the second process target picture block is located, or the CTB where the candidate reference block is located is the second process target picture block. If is not a left-adjacent or right-adjacent block of the CTB in which it is located, it indicates that there is sufficient time to complete the update process for the candidate reference block. Furthermore, since the final motion vector is the motion vector obtained after the initial motion vector has been refined, it is not possible to use the final motion vector of the candidate reference block as the motion vector of the candidate reference block. , Decryption quality can be guaranteed.

2003: The decoder sets a list of predicted motion vectors of candidates based on one or more motion vectors of one or more candidate reference blocks.

Step 2003 is the same as step 1903, i.e., the way the decoder sets the list of candidate predicted motion vectors is consistent with the way the encoder sets the list of candidate predicted motion vectors. , The list of predicted motion vectors of the set candidates is the same. The encoder is an encoder that encodes the bitstream received by the decoder.

2004: The decoder determines the reference block of the picture block to be processed and the motion vector of the reference block from the list of predicted motion vectors of the candidates based on the second identification information.

The motion vector indicated by the index of the second identification information is selected from the list of candidate predicted motion vectors based on the index. The motion vector is the initial motion vector of the second processed picture block, and the motion vector is also the motion vector of the reference block of the second processed picture block.

2005: The encoder stores the motion vector of the reference block as the initial motion vector of the second processed picture block.

Steps 2001 to 2005 are processes in which the decoder acquires the initial motion vector of the picture block to be processed.

2006: The decoder determines the predictive block of the second processed picture block based on the initial motion vector of the second processed picture block and one or more preset motion vector offsets. decide.

Step 2006 can be completed by using 14021 to 140 2 4 in FIG. Details will not be explained here again.

See step 1906 for one or more preset motion vector offsets. Details will not be explained here again.

It should be noted that there are unspecified sequences that perform steps 2005 and 2006, and step 2006 may be performed before step 2005 instead. The sequences that perform steps 2005 and 2006 are not specifically limited in this embodiment of the present application.

2007: After determining the motion vector of the predicted block, the decoder uses the motion vector of the predicted block as the final motion vector of the picture block to be processed.

In step 2007, the motion vector of the candidate reference block can be determined based on the pre-set range of the predicted block. For example, if the predicted block is within the preset range, the initial motion vector of the predicted block is used as the motion vector of the predicted block, and if the predicted block is outside the preset range, the prediction is made. The final motion vector of the block is used as the motion vector of the predicted block.

Specifically, the preset range may be the CTB row range. For example, if the CTB where the prediction block is located and the CTB where the second picture block to be processed is located are located on the same row, the initial motion vector of the prediction block is used as the motion vector of the prediction block. Will be done. Correspondingly, if the CTB where the prediction block is located and the CTB where the second picture block to be processed is located are not located in the same row, the initial motion vector of the prediction block is the motion vector of the prediction block. Used as. Referring to FIG. 21, the coding tree block in which the prediction block is located and the coding tree block in which the second processing target picture block is located are located on different rows, and the prediction block is located in the coding tree block. If the tree block and the coding tree block where the second processed picture block is located are located in adjacent spaces on different rows, the initial motion vector of the predictor block is the motion of the predictor block. It will be understood that it is used as a vector. The adjacent space of different rows may be the upper space, upper left space, etc. of different rows. The particular range of adjacent spaces in different rows is not specifically limited in this embodiment of the invention.

If the CTB where the prediction block is located and the CTB where the second picture block to be processed is located are located on the same row, the update processing for the prediction block may not be completed. Therefore, the initial motion vector of the predicted block can be used directly as the motion vector of the predicted block, and it is not necessary to wait until the predicted block is updated, which reduces the decoding delay. If the CTB where the predicted block is located and the CTB where the second processed picture block is located are not located on the same row, it has enough time to complete the update process for the predicted block. Show that. Therefore, the prediction block has a final motion vector, and the final motion vector of the prediction block can be used as the motion vector of the prediction block. Furthermore, since the final motion vector is the motion vector obtained after the initial motion vector has been refined, using the final motion vector as the motion vector of the prediction block ensures decoding quality. Can be done.

In some embodiments, the preset range may be the CTB block range instead. In some other embodiments, the preset range may be the same CTB block range instead. Determining the motion vector of the predictive block is similar to determining the motion vector of the candidate reference block. See step 1802 for this. Details will not be explained here again.

Steps 2006 and 2007 are processes in which the decoder acquires the final motion vector of the picture block to be processed.

2008: The motion vector residual of the picture block to be processed second is stored in the target storage space.

The target storage space is a space for storing motion vector residuals by the decoder. For example, the target storage space may be the reference picture storage 92 of FIG.

It should be noted that in merge mode in related technology, the target storage space stores zero vectors or the target storage space does not store data. However, in the merge mode of this embodiment of the present application, the motion vector residuals of the picture block to be processed are stored in the target storage space, and as a result, the target storage space can be completely used. Not only that, it is also possible to flexibly select the motion vector of the other picture block when the motion vector of the other picture block is required for the second processed picture block. .. As shown in steps 2002 and 2007, the decoding efficiency of the decoder is further improved.

Further, in some implementations, in step 2008, the final motion vector of the second processed picture block is instead used as the motion vector residual of the second processed picture block. , That is, the final motion vector of the second processed picture block is stored in the target storage space and is stored in the target storage space, that is, the second processed picture block. The motion vector residuals of are not stored in the target storage space. In this case, if the final motion vector of the second processed picture block is required for another block, the final motion vector of the second processed picture block is the second. Can be obtained directly from the decoder's target storage space without the need to add the initial motion vector of the picture block to be processed and the motion vector residuals of the second picture block to be processed. It is possible.

See FIG. 21 for a specific storage method in this step 2008.

It should be noted that there are no strict requirements for the sequence between steps 2007 and 2008, and the decoder may optionally perform step 2007 before step 2008. This is not specifically limited in this embodiment of the present application.

2009: The decoder acquires the picture information corresponding to the second processed picture block based on the motion vector residual of the second processed picture block and the initial motion difference.

The picture information includes pixel information used to identify the original picture of the second processable picture block.

The decoder acquires the picture information corresponding to the picture block to be processed second based on the motion vector residual of the picture block to be processed and the initial motion difference. See FIG. 3 for this. Details will not be explained here again.

Accordingly, the decoder may instead acquire the picture information corresponding to the second processed picture block based on the final motion vector of the second processed picture block. ..

In the above implementation, when decoding the current decrypted block, the decoder updates the reference block as the predicted motion vector of the current decoded block instead of the final motion vector of the reference block. Use an early motion vector that is not. Thus, if the motion vector of the reference block is required for the current decoded block, the relevant steps are performed without having to wait for the final motion vector of the reference block to be updated. It is possible to ensure that the decoding delay is reduced, while the decoding efficiency is improved as the motion vector is updated.

To further reflect the motion vector storage method, specific embodiments are used herein to illustrate the storage steps. The specific explanation is as follows:

In the merge mode, there may be two MVs in each prediction direction, the first MV information and the second MV information, respectively. For example, the first MV and the second MV may be the initial motion vector of the reference block and the final motion vector of the candidate of the reference block, respectively. The MVD derivation process is introduced into merge mode. For example, in the forward prediction process, if the first MV is MV0 and the second MV is MV0', then there is a derivative relationship: MV0'= MV0 + (-) MVD0. Two MVs, eg, the initial motion vector. By storing the index value of and the previous MVD, i.e., the difference between the two MVs, the first MV can be used under certain conditions and the second MV is used under certain conditions. It is possible to avoid delays.

See step 1902 or 2002 for specific conditions. Details will not be explained here again.

In a feasible implementation
In non-merged mode, the target storage space is used to store the motion vector residuals of the reference block, and the motion vector residuals of the reference block are predicted for the final motion vector of the reference block and the reference block. It is the difference from the motion vector.
In merge mode, the target storage space is used to store the selected final motion vector, or the target storage space is the initial motion vector of the reference block and the selected final motion vector. Used to store the difference to and from the motion vector.

Another embodiment of the present application is:

As shown in FIG. 22, the current coding block is the first coding block, which provides the predicted motion information of the current coding block, the reference block is in merge mode, and the motion information of the reference block. The position of is not in the same CTB row range as the current coding block. In this case, the forward motion vector predictor (-21,18) of the current coding block is obtained by adding the forward motion vector (-22,18) and the forward MVD (1,0) of the reference block. The backward motion vector predictor (1,12) of the current coding block is obtained by adding the backward motion vector (2,12) and the backward MVD (-1,0) of the reference block and is the current coding block. The POC of the picture in which the block is located is 4, the POC of the forward reference picture indicated by the index value of the reference picture is 2, and the POC of the backward reference picture indicated by the index value of the reference picture is 6. In this case, the POC corresponding to the current coding block is 4, the POC corresponding to the forward predictive reference picture block is 2, and the POC corresponding to the backward predictive reference picture block is 6.

(-21, 18) are used as a reference input for the forward motion vector predictor, and a first-precision motion search is performed on the forward predictive reference picture block. In this case, the first precision is one pixel precision in the one pixel range. The first forward prediction block is used as a reference and the corresponding new forward prediction block is acquired by each motion search. (1, 12) is used as a reference input for the backward motion vector predictor, and a first-precision motion search is performed on the backward predictive reference picture block. In this case, the first precision is one pixel precision in the one pixel range. The first backward prediction block is used as a reference and the corresponding new backward prediction block is acquired by each motion search. The new forward prediction block is compared with the new backward prediction block to get the forward prediction block with the minimum difference and the backward prediction block with the minimum difference, and into the forward motion vector predictor and the backward prediction block corresponding to the forward prediction block. The corresponding backward motion vector predictors are used as target motion vector predictors and are assumed to be (-21,19) and (1,11), respectively.

The forward MVD (0,1) of the current coding block is obtained by subtracting the forward motion vector predictor (-21,18) from the forward target motion vector predictor (-21,19) and is the current The backward MVD (0, -1) of the coding block is obtained by subtracting the backward motion vector predictor (1,12) from the backward target motion vector predictor (1,11). The MVD calculation method 1 is as follows: MVD0 = (-21,19) − (-21,18) and MVD1 = (1,11) − (1,12); or MVD0 = (-21,19). − (-21, 18) and MVD1 = −MVD0.

Further, forward and backward predictions are performed separately for the first coding block based on the target motion vector predictor, and the obtained forward and backward prediction blocks are used as the target decoding prediction block. The prediction block of the current coding block is updated.

If the first precision motion search is performed on the forward predictive reference picture block and the backward predictive reference picture block, the first precision is any specified precision, eg, integer pixel precision, 1/2. It may be pixel precision, 1/4 pixel precision, or 1/8 pixel precision.

Another embodiment of the present application is:

As shown in FIG. 22, the current coding block is the first coding block, which provides the predicted motion information of the current coding block, the reference block is in merge mode, and the motion information of the reference block. The position of is not in the same CTB range as the current coding block. In this case, the forward motion vector predictor (-21,18) of the current coding block is obtained by adding the forward motion vector (-22,18) and the forward MVD (1,0) of the reference block. The backward motion vector predictor (1,12) of the current coding block is obtained by adding the backward motion vector (2,12) and the backward MVD (-1,0) of the reference block and is the current coding block. The POC of the picture in which the block is located is 4, the POC of the forward reference picture indicated by the index value of the reference picture is 2, and the POC of the backward reference picture indicated by the index value of the reference picture is 6. In this case, the POC corresponding to the current coding block is 4, the POC corresponding to the forward predictive reference picture block is 2, and the POC corresponding to the backward predictive reference picture block is 6.

(-21, 18) are used as a reference input for the forward motion vector predictor, and a first-precision motion search is performed on the forward predictive reference picture block. In this case, the first precision is one pixel precision in the one pixel range. The first forward prediction block is used as a reference and the corresponding new forward prediction block is acquired by each motion search. (1, 12) is used as a reference input for the backward motion vector predictor, and a first-precision motion search is performed on the backward predictive reference picture block. In this case, the first precision is one pixel precision in the one pixel range. The first backward prediction block is used as a reference and the corresponding new backward prediction block is acquired by each motion search.
The new forward prediction block is compared with the new backward prediction block to get the forward prediction block with the minimum difference and the backward prediction block with the minimum difference, and into the forward motion vector predictor and the backward prediction block corresponding to the forward prediction block. The corresponding backward motion vector predictors are used as target motion vector predictors and are assumed to be (-21,19) and (1,11), respectively.

The forward MVD (0,1) of the current coding block is obtained by subtracting the forward motion vector predictor (-21,18) from the forward target motion vector predictor (-21,19) and is the current The backward MVD (0, -1) of the coding block is obtained by subtracting the backward motion vector predictor (1,12) from the backward target motion vector predictor (1,11). The MVD calculation method 1 is as follows: MVD0 = (-21,19) − (-21,18) and MVD1 = (1,11) − (1,12); or MVD0 = (-21,19). − (-21, 18) and MVD1 = −MVD0.

Further, forward and backward predictions are performed separately for the first coding block based on the target motion vector predictor, and the obtained forward and backward prediction blocks are used as the target decoding prediction block. The prediction block of the current coding block is updated.

If the first precision motion search is performed on the forward predictive reference picture block and the backward predictive reference picture block, the first precision is any specified precision, eg, integer pixel precision, 1/2. It may be pixel precision, 1/4 pixel precision, or 1/8 pixel precision.

Another embodiment of the present application is:

As shown in FIG. 22, the current coding block is the first coding block, which provides the predicted motion information of the current coding block, the reference block is in merge mode, and the motion information of the reference block. The position of is not in the same CTB line range as the current coding block. In this case, the forward motion vector predictor (-21,18) of the current coding block is obtained by adding the forward motion vector (-22,18) and the forward MVD (1,0) of the reference block. The backward motion vector predictor (1,12) of the current coding block is obtained by adding the backward motion vector (2,12) and the backward MVD (-1,0) of the reference block and is the current coding block. The POC of the picture in which the block is located is 4, the POC of the forward reference picture indicated by the index value of the reference picture is 2, and the POC of the backward reference picture indicated by the index value of the reference picture is 6. In this case, the POC corresponding to the current coding block is 4, the POC corresponding to the forward predictive reference picture block is 2, and the POC corresponding to the backward predictive reference picture block is 6.

(-21, 18) are used as a reference input for the forward motion vector predictor, and a first-precision motion search is performed on the forward predictive reference picture block. In this case, the first precision is one pixel precision in the one pixel range. The first forward prediction block is used as a reference and the corresponding new forward prediction block is acquired by each motion search. (1, 12) is used as a reference input for the backward motion vector predictor, and a first-precision motion search is performed on the backward predictive reference picture block. In this case, the first precision is one pixel precision in the one pixel range. The first backward prediction block is used as a reference and the corresponding new backward prediction block is acquired by each motion search. The new forward prediction block is compared with the new backward prediction block to get the forward prediction block with the minimum difference and the backward prediction block with the minimum difference, and into the forward motion vector predictor and the backward prediction block corresponding to the forward prediction block. The corresponding backward motion vector predictors are used as target motion vector predictors and are assumed to be (-21,19) and (1,11), respectively.

(-21,19) is then used as a reference input for the forward motion vector predictor, and a first precision motion search is performed on the forward predictive reference picture block. In this case, the first precision is 1/2 pixel precision. The first forward prediction block is used as a reference and the corresponding new forward prediction block is acquired by each motion search. (1, 11) is used as a reference input for the backward motion vector predictor, and a first-precision motion search is performed on the backward predictive reference picture block. In this case, the first precision is 1/2 pixel precision. The first backward prediction block is used as a reference and the corresponding new backward prediction block is acquired by each motion search. The new forward prediction block is compared with the new backward prediction block to get the forward prediction block with the minimum difference and the backward prediction block with the minimum difference.
The forward motion vector predictor corresponding to the forward prediction block and the backward motion vector predictor corresponding to the backward prediction block are used as the target motion vector predictor.

The forward MVD (0,1) of the current coding block is obtained by subtracting the forward motion vector predictor (-21,18) from the forward target motion vector predictor (-21,19) and is the current The backward MVD (0, -1) of the coding block is obtained by subtracting the backward motion vector predictor (1,12) from the backward target motion vector predictor (1,11). The MVD calculation method 1 is as follows: MVD0 = (-21,19) − (-21,18) and MVD1 = (1,11) − (1,12); or MVD0 = (-21,19). − (-21, 18) and MVD1 = −MVD0.

Further, forward and backward predictions are performed separately for the first coding block based on the target motion vector predictor, and the obtained forward and backward prediction blocks are used as the target decoding prediction block. The prediction block of the current coding block is updated.

If the first precision motion search is performed on the forward predictive reference picture block and the backward predictive reference picture block, the first precision is any specified precision, eg, integer pixel precision, 1/2. It may be pixel precision, 1/4 pixel precision, or 1/8 pixel precision.

Another embodiment of the present application is:

The current coding block is the first coding block, and the predicted motion information of the current coding block is acquired. In this case, the forward motion vector predictor (-21,18) of the current coding block is the forward motion vector (-21,18) of the reference block and the backward motion vector predictor (1) of the current coding block. , 12) are the backward movement vectors (1,12) of the reference block, the POC of the picture in which the current coding block is located is 4, and the POC of the forward reference picture indicated by the index value of the reference picture is 2. The POC of the back-referenced picture indicated by the index value of the reference picture is 6. In this case, the POC corresponding to the current coding block is 4, the POC corresponding to the forward predictive reference picture block is 2, and the POC corresponding to the backward predictive reference picture block is 6.

(-21, 18) are used as a reference input for the forward motion vector predictor, and a first-precision motion search is performed on the forward predictive reference picture block. In this case, the first precision is one pixel precision in the one pixel range. The first forward prediction block is used as a reference and the corresponding new forward prediction block is acquired by each motion search. (1, 12) is used as a reference input for the backward motion vector predictor, and a first-precision motion search is performed on the backward predictive reference picture block. In this case, the first precision is one pixel precision in the one pixel range. The first backward prediction block is used as a reference and the corresponding new backward prediction block is acquired by each motion search. The new forward prediction block is compared with the new backward prediction block to get the forward prediction block with the minimum difference and the backward prediction block with the minimum difference, and into the forward motion vector predictor and the backward prediction block corresponding to the forward prediction block. The corresponding backward motion vector predictors are used as target motion vector predictors and are assumed to be (-21,19) and (1,11), respectively.

The forward MVD (0,1) of the current coding block is obtained by subtracting the forward motion vector predictor (-21,18) from the forward target motion vector predictor (-21,19) and is the current The backward MVD (0, -1) of the coding block is obtained by subtracting the backward motion vector predictor (1,12) from the backward target motion vector predictor (1,11). The MVD calculation method 1 is as follows: MVD0 = (-21,19) − (-21,18) and MVD1 = (1,11) − (1,12); or MVD0 = (-21,19). − (-21, 18) and MVD1 = −MVD0.

Further, forward and backward predictions are performed separately for the first coding block based on the target motion vector predictor, and the obtained forward and backward prediction blocks are used as the target decoding prediction block. The prediction block of the current coding block is updated.

If the first precision motion search is performed on the forward predictive reference picture block and the backward predictive reference picture block, the first precision is any specified precision, eg, integer pixel precision, 1/2. It may be pixel precision, 1/4 pixel precision, or 1/8 pixel precision.

In some embodiments, the picture block to be processed may contain a plurality of sub-blocks, and the codec is the final of the picture blocks to be processed based on the final motion vector of each sub-block. Motion vector can be obtained. In a possible implementation, each sub-block has an initial motion vector and a plurality of pre-set offset vectors of the sub-block. The codec has an initial motion vector of any one of a plurality of sub-blocks and a plurality of pre-set offset vectors in order to obtain the final motion vector of multiple candidates of any sub-block. Can be added separately. The codec uses, as the final motion vector of any sub-block, the candidate final motion vector corresponding to the minimum distortion cost in the final motion vector of multiple candidates of any sub-block. Similarly, the codec can obtain the final motion vector of each sub-block, and then the codec can obtain the final motion vector of the picture block to be processed based on the final motion vector of each sub-block. Motion vector can be obtained. The process of obtaining the final motion vector of the picture block to be processed based on the final motion vector of each sub-block is not specifically limited in this embodiment of the present invention.

Before obtaining the final motion vector of multiple candidates for any subblock, the codec must first determine the initial motion vector of any subblock and the preset offset vectors. It should be noted that there is. The method of determining the initial motion vector of any sub-block is similar to the method of determining the initial motion vector of the picture block to be processed and is not described in this embodiment of the invention. Furthermore, the method of determining multiple pre-set offset vectors of any sub-block is similar to the method of determining multiple pre-set offset vectors of the picture block to be processed. Not described in this embodiment of the invention.

FIG. 23 is a schematic block diagram of the motion vector acquisition device 2300 according to the embodiment of the present application. Device 2300 is:
A decision module 2301 configured to determine a reference block for a picture block to be processed, wherein the reference block and the picture block to be processed have a preset temporal or spatial correlation and are referenced. The block has an initial motion vector and one or more preset motion vector offsets, and the initial motion vector of the reference block is obtained based on the predicted motion vector of the reference block and is a reference block. The prediction block of is obtained based on the initial motion vector and one or more pre-set motion vector offsets, with a decision module,
Includes acquisition module 2302 configured to use the initial motion vector of the reference block as the predicted motion vector of the picture block to be processed.

In a feasible implementation, the acquisition module 2302 is configured to use the predicted motion vector of the reference block as the initial motion vector of the reference block, or the predicted motion vector of the reference block and the reference block. It is further configured to obtain the initial motion vector of the reference block by adding the motion vector difference of.

In a feasible implementation, acquisition module 2302 further acquires the picture block represented by the initial motion vector of the reference block from the reference frame of the reference block so that it acts as a temporal prediction block of the reference block. To get one or more actual motion vectors, add the initial motion vector of the reference vector and one or more preset motion vector offsets (where each actual motion vector is Obtain one or more candidate prediction blocks at one or more search positions indicated by one or more actual motion vectors (indicating one or more search positions) (where each search position is one). A candidate prediction block having the smallest pixel difference from the temporal prediction block (corresponding to the candidate prediction block) is configured to be selected from one or more candidate prediction blocks as the reference block prediction block. ..

In a feasible implementation, the device 2300 is used for bidirectional prediction, the reference frame includes a reference frame in the first direction and a reference frame in the second direction, and the initial motion vector is the initial motion vector in the first direction. It contains a motion vector and an initial motion vector in the second direction, and the acquisition module 2302 is specifically indicated by the initial motion vector in the first direction of the reference block from the reference frame in the first direction of the reference block. Obtain the first picture block, and from the reference frame in the second direction of the reference block, obtain the second picture block indicated by the initial motion vector in the second direction of the reference block, and obtain the temporal of the reference block. It is configured to perform a weighting process on the first picture block and the second picture block in order to acquire the prediction block.

In a feasible implementation, acquisition module 2302 is specifically configured to use the predicted motion vector of the picture block to be processed as the initial motion vector of the picture block to be processed.

In a feasible implementation, acquisition module 2302 adds the predicted motion vector of the picture block to be processed and the motion vector difference of the picture block to be processed to the initial of the picture block to be processed. It is specifically configured to get a motion vector.

In a feasible implementation, the device 2300 is used for video decoding and the motion vector differences of the picture blocks to be processed are obtained by analyzing the first discriminant information in the bitstream.

In a feasible implementation, device 2300 is used for video decoding and decision module 2301 analyzes the bitstream to obtain second identification information and is based on the second identification information of the picture block to be processed. It is specifically configured to determine the reference block.

In a feasible implementation, device 2300 is used for video coding and the decision module 2301 is taken from one or more candidate reference blocks of the picture block to be processed as a reference block of the picture block to be processed. It is specifically configured to select the candidate reference block with the minimum rate distortion cost.

FIG. 24 is a schematic block diagram of a video coding device according to an embodiment of the present application. The device 2400 may be applied to the encoder or may be applied to the decoder. Device 2400 includes processor 2401 and memory 2402. The processor 2401 and the memory 2402 are connected to each other (eg, via bus 2404). In a possible implementation, the device 2400 may further include a transceiver 2403. Transceiver 2403 is connected to processor 2401 and memory 2402 and is configured to receive / transmit data.

The memory 2402 is a random access memory (RAM), a read-only memory (ROM), and an erasable programmable. Includes, but is not limited to, read-only memory (EPROM) or compact disc read-only memory (CD-ROM). Memory 2402 is configured to store relevant program code and video data.

Processor 2401 may be one or more central processing units (CPUs). When the processor 2401 is one CPU, the CPU may be a single core CPU or a multi-core CPU.

Processor 2401 is configured to read the program code stored in memory 2402 and perform any implementation solution corresponding to FIGS. 13 to 20 and operations in various feasible implementations of the implementation solution. To.

For example, embodiments of the present application further provide a computer-readable storage medium. A computer-readable storage medium stores instructions. When the instruction is executed on the computer, the computer
It makes it possible to perform operations in any implementation solution corresponding to FIGS. 13 to 20 and in various implementable implementations of the implementation solution.

For example, embodiments of the present application further provide computer program products that include instructions. When a computer program product is run on a computer, the computer is capable of performing any implementation solution corresponding to FIGS. 13-20, as well as operations in various implementable implementations of the implementation solution. To.

FIG. 25 shows a motion vector acquisition device according to an embodiment of the present application. The apparatus includes a determination module 2501, a first acquisition module 2502, and a second acquisition module 2503.

The decision module 2501 is configured to perform step 1904.

The first acquisition module 2502 is configured to perform step 1907.

The second acquisition module 2503 is configured to perform step 1907.

Optionally, the reference block is in the preset range so that the coding tree block CTB where the reference block is located and the coding tree block where the picture block to be processed is located are on the same line. Including being located.
Correspondingly, the fact that the reference block is out of the preset range is related to step 1907, that is, the coding tree block in which the reference block is located and the coding tree in which the picture block to be processed is located. -Includes that the block is located on a different line.

Optionally, having the reference block in the preset range includes that the reference block and the picture block to be processed are in the same coding tree block.
Correspondingly, the fact that the reference block is out of the preset range includes that the reference block and the picture block to be processed are located in different coding tree blocks in relation to step 1907.

As an option
The fact that the reference block is within the preset range means that the coding tree block in which the reference block is located is the same as the coding tree block in which the picture block to be processed is located, or the reference block is located. Includes that the located coding tree block is a left-adjacent or right-adjacent block of the coding tree block in which the picture block to be processed is located.
Correspondingly, the fact that the reference block is out of the preset range means that for step 1907, the coding tree block in which the reference block is located is the same as the coding tree block in which the picture block to be processed is located. Includes that the coding tree block in which the reference block is located is not the left-adjacent or right-adjacent block of the coding tree block in which the picture block to be processed is located.

FIG. 26 shows a motion vector residual determination device according to an embodiment of the present application. The apparatus includes an analysis module 2601, an addition module 2602, a determination module 2603, and an acquisition module 2604.

The analysis module 2601 is configured to perform step 2001.

Addition module 2602 is configured to perform step 2002.

The decision module 2603 is configured to perform step 1422.

Acquisition module 2604 is configured to perform step 2008.

Optionally, the device is used for bidirectional inter-prediction, the final motion vector contains a first final motion vector and a second final motion vector, and the initial motion vector is the first initial motion vector. Includes motion vector and second initial motion vector; first final motion vector and first initial motion vector are motion compensation blocks based on the first reference frame list of the picture block to be processed. The second final motion vector and the second initial motion vector indicate the motion compensation block based on the second reference frame list of the picture block to be processed; and the final motion vector and Using the difference from the initial motion vector as the motion vector residuals of the picture block being processed is:
The difference between the first final motion vector and the first initial motion vector is used as the first motion vector residual of the picture block to be processed.

FIG. 27 shows a motion vector data storage device according to an embodiment of the present application. The apparatus includes a first analysis module 2701, an acquisition module 2702, a second analysis module 2703, and a storage module 2704.

The first analysis module 2701 is configured to perform step 2001.

Acquisition module 2702 is configured to perform step 2007.

The second analysis module 2703 is configured to perform step 2001.

Storage module 2704 is configured to perform step 2008.

Those skilled in the art will appreciate that in combination with the examples described in the embodiments disclosed herein, the units and algorithm steps are implemented by electronic hardware, or a combination of computer software and electronic hardware. Will recognize that is possible. Whether a function is performed by hardware or software depends on the design constraints of the particular application and technical solution. One of ordinary skill in the art can use various methods to implement the described functionality for each particular application, but such implementation should be considered to be beyond the scope of this application. No.

It will be apparent to those skilled in the art that for convenience and brief description purposes, those skilled in the art should refer to the corresponding processes in embodiments of the methods described above for detailed operating processes of the systems, devices and units described above. Will be done. Details will not be explained here again.

All or part of the aforementioned embodiments can be realized by software, hardware, firmware, or any combination thereof. When the software is used to implement an embodiment, the embodiment can be fully or partially implemented in the form of a computer program product. Computer program products include one or more computer instructions. When computer program instructions are loaded and executed on a computer, all or part of the procedure or function arises according to embodiments of the present application. The computer may be a general purpose computer, a dedicated computer, a computer network, or other programmable device. Computer instructions may be stored on a computer-readable storage medium or may be transmitted from a computer-readable storage medium to another computer-readable storage medium. For example, a computer command can be wired (eg, coaxial cable, fiber optic, or digital subscriber line) or from a website, computer, server, or data center to another website, computer, server, or data center. It can be transmitted wirelessly (eg, infrared, microwave, etc.). The computer-readable storage medium may be any usable medium accessible by the computer, or a data storage device such as a server or data center that integrates one or more usable media. The medium that can be used may be a magnetic medium (for example, a floppy disk, a hard disk, or a magnetic tape), an optical medium (for example, a digital versatile disk), a semiconductor medium (for example, a solid-state drive), or the like.

In the aforementioned embodiments, the description in each embodiment has an individual focus. For parts not described in detail in the embodiment, refer to the related description in the other embodiments.

The above description merely describes a particular implementation of the present application and is not intended to limit the scope of protection of the present application. Any modifications or substitutions readily apparent to those skilled in the art within the technical scope disclosed herein shall be within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be in accordance with the scope of protection of the claim.

Claims (32)

It is a motion vector acquisition method.
A step of determining a reference block of a picture block to be processed, wherein the reference block and the picture block to be processed are located in the same frame of the picture.
A step of acquiring the motion vector of the picture block to be processed based on the initial motion vector of the reference block when the reference block is within the preset range, which is set in advance. The range is determined based on the position of the picture block to be processed.
A step of acquiring the motion vector of the picture block to be processed based on the final motion vector of the reference block when the reference block is outside the preset range. Motion vector is obtained based on the initial motion vector, including steps. The fact that the reference block is within the preset range means that the coding tree block CTB where the reference block is located and the coding tree block where the picture block to be processed is located are located on the same row. Including doing
Correspondingly, the fact that the reference block is out of the preset range means that the coding tree block in which the reference block is located and the coding tree block in which the picture block to be processed is located. The method of claim 1, wherein the method comprises being located on a different line. The coding tree block in which the reference block is located and the coding tree block in which the picture block to be processed is located are located on different lines.
The method of claim 2, wherein the coding tree block in which the reference block is located is above or in the upper left of the coding tree block in which the picture block to be processed is located. The fact that the reference block is within the preset range includes that the reference block and the picture block to be processed are located in the same coding tree block.
Accordingly, the fact that the reference block is out of the preset range includes that the reference block and the picture block to be processed are located in different coding tree blocks. The method described in. The fact that the reference block is within the preset range means that the coding tree block in which the reference block is located is the same as the coding tree block in which the picture block to be processed is located, or The coding tree block in which the reference block is located comprises being a left-adjacent or right-adjacent block of the coding tree block in which the picture block to be processed is located, and correspondingly, the reference block is Being out of the preset range means that the coding tree block in which the reference block is located is not the same as the coding tree block in which the picture block to be processed is located, or said. The method of claim 1, wherein the coding tree block in which the reference block is located is not a left-adjacent or right-adjacent block of the coding tree block in which the picture block to be processed is located. The step of determining the reference block of the picture block to be processed is
A step of continuously determining one or more previously set candidate reference blocks as the reference blocks in a preset order, wherein the candidate reference blocks are set in advance with the picture block to be processed. The method of any one of claims 1-5, comprising a step, comprising a picture block having a spatially positional relationship. The step of determining the reference block of the picture block to be processed is
The step of parsing the bitstream to obtain one or more first identification information,
A step of determining the reference block from a plurality of candidate reference blocks of the picture block to be processed based on the one or more first identification information, wherein the candidate reference block is a picture to be processed. The method of any one of claims 1-5, comprising a step and comprising a picture block having a spatially positional relationship set in advance with the block. It is possible that the final motion vector is obtained based on the initial motion vector.
To obtain multiple candidates for the final motion vector by adding the initial motion vector and multiple preset offset vectors separately.
Of claims 1-7, which comprises determining the final motion vector candidate corresponding to the smallest strain cost among the plurality of final motion vector candidates as the final motion vector. The method according to any one of the above. The method is used for bidirectional inter-prediction, the final motion vector includes a first final motion vector and a second final motion vector, and the initial motion vector is the first initial motion vector. A motion vector and a second initial motion vector are included, and the first final motion vector and the first initial motion vector are included in the first reference frame list of the picture block to be processed. The motion compensation block is shown based on the above, and the second final motion vector and the second initial motion vector are motion compensation blocks based on the second reference frame list of the picture block to be processed. Show,
It is possible that the final motion vector is obtained based on the initial motion vector.
To obtain multiple candidates for the first final motion vector by separately adding the first initial motion vector and the plurality of pre-set offset vectors.
The candidate for the first final motion vector corresponding to the minimum strain cost among the plurality of candidates for the first final motion vector is determined as the first final motion vector. The first final motion vector corresponds to the first offset vector in the plurality of previously set offset vectors.
By acquiring the second offset vector, the size of the second offset vector is equal to that of the first offset vector, and the direction of the second offset vector is that of the first offset vector. And the opposite,
The first aspect of claim 1-7, which comprises adding the second initial motion vector and the second offset vector to obtain the second final motion vector. the method of. The method is used for bidirectional inter-prediction, the final motion vector includes a first final motion vector and a second final motion vector, and the initial motion vector is the first initial motion vector. A motion vector and a second initial motion vector are included, and the first final motion vector and the first initial motion vector are included in the first reference frame list of the picture block to be processed. The motion compensation block is shown based on the above, and the second final motion vector and the second initial motion vector are motion compensation blocks based on the second reference frame list of the picture block to be processed. Show,
It is possible that the final motion vector is obtained based on the initial motion vector.
To obtain multiple candidates for the first final motion vector by separately adding the first initial motion vector and the plurality of pre-set offset vectors.
Determining the candidate for the first final motion vector corresponding to the minimum strain cost among the plurality of candidates for the first final motion vector as the first final motion vector.
To obtain multiple candidates for the second final motion vector by separately adding the second initial motion vector and the plurality of pre-set second offset vectors.
The candidate for the second final motion vector corresponding to the minimum strain cost among the plurality of candidates for the second final motion vector is determined as the second final motion vector. The method according to any one of claims 1-7, including. It is a motion vector residual determination method.
The step of analyzing the bitstream to obtain the second identification information, wherein the second identification information is used to determine the initial motion vector of the picture block to be processed.
The step of adding the initial motion vector and the plurality of pre-set offset vectors separately to obtain the final motion vector of a plurality of candidates.
A step of determining the final motion vector of the candidate corresponding to the minimum strain cost among the final motion vectors of the plurality of candidates as the final motion vector, and
The difference between the final motion vector and the initial motion vector is used as the motion vector residual of the picture block to be processed, or the final motion vector is used as the picture to be processed. • A method that includes steps and steps to use as the motion vector residuals of the block. The method is used for bidirectional inter-prediction, the final motion vector includes a first final motion vector and a second final motion vector, and the initial motion vector is the first initial motion vector. A motion vector and a second initial motion vector are included, and the first final motion vector and the first initial motion vector are included in the first reference frame list of the picture block to be processed. The motion compensation block is shown based on the above, and the second final motion vector and the second initial motion vector are motion compensation blocks based on the second reference frame list of the picture block to be processed. Show,
The step of using the difference between the final motion vector and the initial motion vector as the motion vector residual of the picture block to be processed is
11. The eleventh claim comprises the step of using the difference between the first final motion vector and the first initial motion vector as the first motion vector residual of the picture block to be processed. The method described. The step of using the difference between the final motion vector and the initial motion vector as the motion vector residuals of the picture block to be processed
12. The step further comprises using the difference between the second final motion vector and the second initial motion vector as the second motion vector residual of the picture block to be processed. The method described in. The step of using the final motion vector as the motion vector residuals of the picture block to be processed
12. The method of claim 12, comprising the step of using the first final motion vector as the first motion vector residual of the picture block to be processed. The step of using the final motion vector as the motion vector residuals of the picture block to be processed
14. The method of claim 14, further comprising the step of using the second final motion vector as the second motion vector residual of the picture block to be processed. It is a motion vector data storage method.
The step of analyzing the bitstream to obtain the second identification information and the third identification information, wherein the second identification information is used to determine an initial predicted motion vector of the picture block to be processed. Used, steps and
A step to obtain the final predicted motion vector based on the initial predicted motion vector and a plurality of pre-set offset vectors.
When the third identification information indicates that the bitstream carries the motion vector residual of the picture block to be processed, the bitstream is analyzed to obtain the motion vector residual. The step of storing the motion vector residual in the target storage space,
The final predicted motion vector and the initial predicted motion vector when the third identification information indicates that the bitstream does not carry the motion vector residuals of the picture block to be processed. A method comprising storing the difference from a motion vector in the target storage space or storing the final predicted motion vector in the target storage space. The third identification information indicates that the bitstream carries the motion vector residuals of the picture block to be processed.
The third identification information indicates that the prediction mode of the picture block to be processed is the AMVP mode.
16. The method of claim 16. The third identification information indicates that the bitstream does not carry the motion vector residuals of the picture block to be processed.
The third identification information indicates that the prediction mode of the picture block to be processed is the merge mode or the skip mode.
16. The method of claim 16 or 17. The step of obtaining the final predicted motion vector based on the initial predicted motion vector and a plurality of pre-set offset vectors is the step.
A step of adding the initial motion vector and a plurality of pre-set offset vectors separately to obtain multiple candidates for the final motion vector.
Of claims 16-18, comprising the step of determining the final motion vector candidate corresponding to the smallest strain cost among the plurality of final motion vector candidates as the final motion vector. The method according to any one of the above. It is a motion vector acquisition device
A decision module configured to determine a reference block of a picture block to be processed, wherein the reference block and the picture block to be processed are located in the same frame of the picture.
In the first acquisition module configured to acquire the motion vector of the picture block to be processed based on the initial motion vector of the reference block when the reference block is within a preset range. Therefore, the preset range is determined based on the position of the picture block to be processed.
A second acquisition configured to acquire the motion vector of the picture block to be processed based on the final motion vector of the reference block when the reference block is outside the preset range. A device including a second acquisition module, which is a module in which the final motion vector is acquired based on the initial motion vector. The fact that the reference block is within the preset range means that the coding tree block CTB where the reference block is located and the coding tree block where the picture block to be processed is located are located on the same row. Including doing
Correspondingly, the fact that the reference block is out of the preset range means that the coding tree block in which the reference block is located and the coding tree block in which the picture block to be processed is located. 20. The apparatus of claim 20, wherein the apparatus is located in a different row. The fact that the reference block is within the preset range includes that the reference block and the picture block to be processed are located in the same coding tree block.
Accordingly, the fact that the reference block is out of the preset range includes that the reference block and the picture block to be processed are located in different coding tree blocks, claim 20. The device described in. The fact that the reference block is within the preset range means that the coding tree block in which the reference block is located is the same as the coding tree block in which the picture block to be processed is located, or The coding tree block in which the reference block is located comprises being a left-adjacent or right-adjacent block of the coding tree block in which the picture block to be processed is located, and correspondingly, the reference block is Being out of the preset range means that the coding tree block in which the reference block is located is not the same as the coding tree block in which the picture block to be processed is located, or said. 20. The apparatus of claim 20, wherein the coding tree block in which the reference block is located is not a left-adjacent or right-adjacent block of the coding tree block in which the picture block to be processed is located. It is a motion vector residual determination device.
An analysis module configured to analyze a bitstream to obtain second identification information, said second identification information used to determine the initial motion vector of the picture block to be processed. , Analysis module,
A total module configured to obtain the final motion vectors of multiple candidates by adding the initial motion vector and multiple preset offset vectors separately.
A determination module configured to determine the final motion vector of the candidate corresponding to the minimum strain cost among the final motion vectors of the plurality of candidates as the final motion vector.
The difference between the final motion vector and the initial motion vector is used as the motion vector residual of the picture block to be processed, or the final motion vector is used as the picture to be processed. • A device that includes a capture module configured to be used as a block motion vector residual. The device is used for bidirectional inter-prediction, the final motion vector includes a first final motion vector and a second final motion vector, and the initial motion vector is the first initial motion vector. A motion vector and a second initial motion vector are included, and the first final motion vector and the first initial motion vector are included in the first reference frame list of the picture block to be processed. The motion compensation block is shown based on the above, and the second final motion vector and the second initial motion vector are motion compensation blocks based on the second reference frame list of the picture block to be processed. Show,
Using the difference between the final motion vector and the initial motion vector as the motion vector residuals of the picture block to be processed can be used.
24. Claim 24 includes using the difference between the first final motion vector and the first initial motion vector as the first motion vector residual of the picture block to be processed. The device described. A motion vector data storage device
A first analysis module configured to analyze a bitstream to obtain second and third identification information, the second identification information being initially predicted for a picture block to be processed. The first analysis module, which is used to determine the motion vector,
An acquisition module configured to acquire the final predicted motion vector based on the initial predicted motion vector and a plurality of pre-set offset vectors.
When the third identification information indicates that the bitstream carries the motion vector residual of the picture block to be processed, the bitstream is analyzed to obtain the motion vector residual. A second analysis module configured to store the motion vector residuals in the target storage space, and
The final predicted motion vector and the initial predicted motion vector when the third identification information indicates that the bitstream does not carry the motion vector residuals of the picture block to be processed. A device comprising a storage module configured to store the difference to and from the motion vector in the target storage space or to store the final predicted motion vector in the target storage space. A computer device, wherein the computer device includes a processor and a memory, the memory stores at least one instruction, and the instruction is the instruction according to any one of claims 1-10. A computer device loaded and executed by the processor to perform the operation performed by the motion vector acquisition method. A computer device, wherein the computer device includes a processor and a memory, the memory stores at least one instruction, and the instruction is the instruction according to any one of claims 11-15. A computer device loaded and executed by the processor to perform the operation performed by the motion vector residual determination method. A computer device, wherein the computer device includes a processor and a memory, the memory stores at least one instruction, and the instruction is the instruction according to any one of claims 16-19. A computer device loaded and executed by the processor to perform the operations performed by the motion vector data storage method. A computer-readable storage medium, wherein the storage medium stores at least one instruction, and the instruction is executed by the motion vector acquisition method according to any one of claims 1-10. A storage medium loaded and executed by the processor to perform the operation. A computer-readable storage medium, wherein the storage medium stores at least one instruction, and the instruction is executed by the motion vector residual determination method according to any one of claims 11-15. A storage medium loaded and executed by the processor to realize the operation to be performed. A computer-readable storage medium, wherein the storage medium stores at least one instruction, which is the motion vector data storage method according to any one of claims 16-19. A storage medium loaded and executed by the processor to realize the operation to be performed.

Images